Skip to content

P
PaddleOCR

PaddlePaddle / PaddleOCR
大约 1 年 前同步成功

通知 1528
Star 32962
Fork 6643
P
PaddleOCR
未验证 提交 a8a9b2e5 编写于 作者: X xiaoting 提交者： GitHub
浏览文件
操作

Merge branch 'dygraph' into fix_prepare

上级 49e42a16 0f5a5d96
展开全部 隐藏空白更改
内联 并排
Showing with 6314 addition and 1193 deletion
README.md
浏览文件 @ a8a9b2e5
English | [简体中文](README_ch.md)
<p align="center">
<img src="./doc/PaddleOCR_log.png" align="middle" width = "600"/>
<p align="center">
------------------------------------------------------------------------------------------
<p align="left">
<a href="./LICENSE"><img src="https://img.shields.io/badge/license-Apache%202-dfd.svg"></a>
<a href="https://github.com/PaddlePaddle/PaddleOCR/releases"><img src="https://img.shields.io/github/v/release/PaddlePaddle/PaddleOCR?color=ffa"></a>
<a href=""><img src="https://img.shields.io/badge/python-3.7+-aff.svg"></a>
<a href=""><img src="https://img.shields.io/badge/os-linux%2C%20win%2C%20mac-pink.svg"></a>
<a href=""><img src="https://img.shields.io/pypi/format/PaddleOCR?color=c77"></a>
<a href="https://github.com/PaddlePaddle/PaddleOCR/graphs/contributors"><img src="https://img.shields.io/github/contributors/PaddlePaddle/PaddleOCR?color=9ea"></a>
<a href="https://pypi.org/project/PaddleOCR/"><img src="https://img.shields.io/pypi/dm/PaddleOCR?color=9cf"></a>
<a href="https://github.com/PaddlePaddle/PaddleOCR/stargazers"><img src="https://img.shields.io/github/stars/PaddlePaddle/PaddleOCR?color=ccf"></a>
</p>
## Introduction
PaddleOCR aims to create multilingual, awesome, leading, and practical OCR tools that help users train better models and apply them into practice.
## Notice
PaddleOCR supports both dynamic graph and static graph programming paradigm
- Dynamic graph: dygraph branch (default), **supported by paddle 2.0.0 ([installation](./doc/doc_en/installation_en.md))**
- Static graph: develop branch
**Recent updates**
- 2021.1.21 update more than 25+ multilingual recognition models [models list](./doc/doc_en/models_list_en.md), including：English, Chinese, German, French, Japanese，Spanish，Portuguese Russia Arabic and so on. Models for more languages will continue to be updated [Develop Plan](https://github.com/PaddlePaddle/PaddleOCR/issues/1048).
- 2020.12.15 update Data synthesis tool, i.e., [Style-Text](./StyleText/README.md)，easy to synthesize a large number of images which are similar to the target scene image.
- 2020.11.25 Update a new data annotation tool, i.e., [PPOCRLabel](./PPOCRLabel/README.md), which is helpful to improve the labeling efficiency. Moreover, the labeling results can be used in training of the PP-OCR system directly.
- 2020.9.22 Update the PP-OCR technical article, https://arxiv.org/abs/2009.09941
- PaddleOCR R&D team would like to share the key points of PP-OCRv2, at 20:15 pm on September 8th, [Live Address](https://live.bilibili.com/21689802).
- 2021.9.7 release PaddleOCR v2.3, [PP-OCRv2](#PP-OCRv2) is proposed. The inference speed of PP-OCRv2 is 220% higher than that of PP-OCR server in CPU device. The F-score of PP-OCRv2 is 7% higher than that of PP-OCR mobile.
- 2021.8.3 released PaddleOCR v2.2, add a new structured documents analysis toolkit, i.e., [PP-Structure](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.2/ppstructure/README.md), support layout analysis and table recognition (One-key to export chart images to Excel files).
- 2021.4.8 release end-to-end text recognition algorithm [PGNet](https://www.aaai.org/AAAI21Papers/AAAI-2885.WangP.pdf) which is published in AAAI 2021. Find tutorial [here](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.1/doc/doc_en/pgnet_en.md)；release multi language recognition [models](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.1/doc/doc_en/multi_languages_en.md), support more than 80 languages recognition; especically, the performance of [English recognition model](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.1/doc/doc_en/models_list_en.md#English) is Optimized.
- [more](./doc/doc_en/update_en.md)
## Features
- PPOCR series of high-quality pre-trained models, comparable to commercial effects
- Ultra lightweight ppocr_mobile series models: detection (3.0M) + direction classifier (1.4M) + recognition (5.0M) = 9.4M
- General ppocr_server series models: detection (47.1M) + direction classifier (1.4M) + recognition (94.9M) = 143.4M
- PP-OCR series of high-quality pre-trained models, comparable to commercial effects
- Ultra lightweight PP-OCRv2 series models: detection (3.1M) + direction classifier (1.4M) + recognition 8.5M) = 13.0M
- Ultra lightweight PP-OCR mobile series models: detection (3.0M) + direction classifier (1.4M) + recognition (5.0M) = 9.4M
- General PP-OCR server series models: detection (47.1M) + direction classifier (1.4M) + recognition (94.9M) = 143.4M
- Support Chinese, English, and digit recognition, vertical text recognition, and long text recognition
- Support multi-language recognition: Korean, Japanese, German, French
- Rich toolkits related to the OCR areas
......@@ -64,39 +82,44 @@ Mobile DEMO experience (based on EasyEdge and Paddle-Lite, supports iOS and Andr
<a name="Supported-Chinese-model-list"></a>
## PP-OCR 2.0 series model list（Update on Dec 15）
**Note** : Compared with [models 1.1](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/doc/doc_en/models_list_en.md), which are trained with static graph programming paradigm, models 2.0 are the dynamic graph trained version and achieve close performance.
## PP-OCR series model list（Update on September 8th）
| Model introduction | Model name | Recommended scene | Detection model | Direction classifier | Recognition model |
| ------------------------------------------------------------ | ---------------------------- | ----------------- | ------------------------------------------------------------ | ------------------------------------------------------------ | ------------------------------------------------------------ |
| Chinese and English ultra-lightweight OCR model (9.4M) | ch_ppocr_mobile_v2.0_xx | Mobile & server |[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_det_infer.tar) / [pre-trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_det_train.tar)|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_infer.tar) / [pre-trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_train.tar) |[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_infer.tar) / [pre-trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_pre.tar) |
| Chinese and English general OCR model (143.4M) | ch_ppocr_server_v2.0_xx | Server |[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_det_infer.tar) / [pre-trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_det_train.tar) |[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_infer.tar) / [pre-trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_traingit.tar) |[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_rec_infer.tar) / [pre-trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_rec_pre.tar) |
| Chinese and English ultra-lightweight PP-OCRv2 model（11.6M） | ch_PP-OCRv2_xx |Mobile&Server|[inference model](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_det_infer.tar) / [pre-trained model](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_det_distill_train.tar)| [inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_infer.tar) / [pre-trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_train.tar) |[inference model](https://paddleocr.bj.bcebos.com/PP-OCRv2/ch/ch_PP-OCRv2_rec_infer.tar) / [pre-trained model](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_rec_train.tar)|
| Chinese and English ultra-lightweight PP-OCR model (9.4M) | ch_ppocr_mobile_v2.0_xx | Mobile & server |[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_det_infer.tar) / [pre-trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_det_train.tar)|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_infer.tar) / [pre-trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_train.tar) |[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_infer.tar) / [pre-trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_pre.tar) |
| Chinese and English general PP-OCR model (143.4M) | ch_ppocr_server_v2.0_xx | Server |[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_det_infer.tar) / [pre-trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_det_train.tar) |[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_infer.tar) / [pre-trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_traingit.tar) |[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_rec_infer.tar) / [pre-trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_rec_pre.tar) |
For more model downloads (including multiple languages), please refer to [PP-OCR v2.0 series model downloads](./doc/doc_en/models_list_en.md).
For more model downloads (including multiple languages), please refer to [PP-OCR series model downloads](./doc/doc_en/models_list_en.md).
For a new language request, please refer to [Guideline for new language_requests](#language_requests).
## Tutorials
- [Installation](./doc/doc_en/installation_en.md)
- [Environment Preparation](./doc/doc_en/environment_en.md)
- [Quick Start](./doc/doc_en/quickstart_en.md)
- [Code Structure](./doc/doc_en/tree_en.md)
- Algorithm Introduction
- [Text Detection Algorithm](./doc/doc_en/algorithm_overview_en.md)
- [Text Recognition Algorithm](./doc/doc_en/algorithm_overview_en.md)
- [PP-OCR Pipeline](#PP-OCR-Pipeline)
- Model Training/Evaluation
- [Text Detection](./doc/doc_en/detection_en.md)
- [Text Recognition](./doc/doc_en/recognition_en.md)
- [Direction Classification](./doc/doc_en/angle_class_en.md)
- [Yml Configuration](./doc/doc_en/config_en.md)
- Inference and Deployment
- [Quick Inference Based on PIP](./doc/doc_en/whl_en.md)
- [PaddleOCR Overview and Installation](./doc/doc_en/paddleOCR_overview_en.md)
- PP-OCR Industry Landing: from Training to Deployment
- [PP-OCR Model and Configuration](./doc/doc_en/models_and_config_en.md)
- [PP-OCR Model Download](./doc/doc_en/models_list_en.md)
- [Yml Configuration](./doc/doc_en/config_en.md)
- [Python Inference for PP-OCR Model Library](./doc/doc_en/inference_ppocr_en.md)
- [PP-OCR Training](./doc/doc_en/training_en.md)
- [Text Detection](./doc/doc_en/detection_en.md)
- [Text Recognition](./doc/doc_en/recognition_en.md)
- [Direction Classification](./doc/doc_en/angle_class_en.md)
- Inference and Deployment
- [C++ Inference](./deploy/cpp_infer/readme_en.md)
- [Serving](./deploy/pdserving/README.md)
- [Mobile](./deploy/lite/readme_en.md)
- [Benchmark](./doc/doc_en/benchmark_en.md)
- [PP-Structure: Information Extraction](./ppstructure/README.md)
- [Layout Parser](./ppstructure/layout/README.md)
- [Table Recognition](./ppstructure/table/README.md)
- Academic Circles
- [Two-stage Algorithm](./doc/doc_en/algorithm_overview_en.md)
- [PGNet Algorithm](./doc/doc_en/algorithm_overview_en.md)
- [Python Inference](./doc/doc_en/inference_en.md)
- [C++ Inference](./deploy/cpp_infer/readme_en.md)
- [Serving](./deploy/pdserving/README.md)
- [Mobile](./deploy/lite/readme_en.md)
- [Benchmark](./doc/doc_en/benchmark_en.md)
- Data Annotation and Synthesis
- [Semi-automatic Annotation Tool: PPOCRLabel](./PPOCRLabel/README.md)
- [Data Synthesis Tool: Style-Text](./StyleText/README.md)
......@@ -114,17 +137,18 @@ For a new language request, please refer to [Guideline for new language_requests
- [License](#LICENSE)
- [Contribution](#CONTRIBUTION)
<a name="PP-OCRv2"></a>
## PP-OCRv2 Pipeline
<div align="center">
<img src="./doc/ppocrv2_framework.jpg" width="800">
</div>
<a name="PP-OCR-Pipeline"></a>
[1] PP-OCR is a practical ultra-lightweight OCR system. It is mainly composed of three parts: DB text detection, detection frame correction and CRNN text recognition. The system adopts 19 effective strategies from 8 aspects including backbone network selection and adjustment, prediction head design, data augmentation, learning rate transformation strategy, regularization parameter selection, pre-training model use, and automatic model tailoring and quantization to optimize and slim down the models of each module (as shown in the green box above). The final results are an ultra-lightweight Chinese and English OCR model with an overall size of 3.5M and a 2.8M English digital OCR model. For more details, please refer to the PP-OCR technical article (https://arxiv.org/abs/2009.09941).
## PP-OCR Pipeline
[2] On the basis of PP-OCR, PP-OCRv2 is further optimized in five aspects. The detection model adopts CML(Collaborative Mutual Learning) knowledge distillation strategy and CopyPaste data expansion strategy. The recognition model adopts LCNet lightweight backbone network, U-DML knowledge distillation strategy and enhanced CTC loss function improvement (as shown in the red box above), which further improves the inference speed and prediction effect. For more details, please refer to the technical report of PP-OCRv2 (arXiv link is coming soon).
<div align="center">
<img src="./doc/ppocr_framework.png" width="800">
</div>
PP-OCR is a practical ultra-lightweight OCR system. It is mainly composed of three parts: DB text detection[2], detection frame correction and CRNN text recognition[7]. The system adopts 19 effective strategies from 8 aspects including backbone network selection and adjustment, prediction head design, data augmentation, learning rate transformation strategy, regularization parameter selection, pre-training model use, and automatic model tailoring and quantization to optimize and slim down the models of each module. The final results are an ultra-lightweight Chinese and English OCR model with an overall size of 3.5M and a 2.8M English digital OCR model. For more details, please refer to the PP-OCR technical article (https://arxiv.org/abs/2009.09941). Besides, The implementation of the FPGM Pruner [8] and PACT quantization [9] is based on [PaddleSlim](https://github.com/PaddlePaddle/PaddleSlim).
## Visualization [more](./doc/doc_en/visualization_en.md)
......
README_ch.md
浏览文件 @ a8a9b2e5
[English](README.md) | 简体中文
<p align="center">
<img src="./doc/PaddleOCR_log.png" align="middle" width = "600"/>
<p align="center">
------------------------------------------------------------------------------------------
<p align="left">
<a href="./LICENSE"><img src="https://img.shields.io/badge/license-Apache%202-dfd.svg"></a>
<a href="https://github.com/PaddlePaddle/PaddleOCR/releases"><img src="https://img.shields.io/github/v/release/PaddlePaddle/PaddleOCR?color=ffa"></a>
<a href=""><img src="https://img.shields.io/badge/python-3.7+-aff.svg"></a>
<a href=""><img src="https://img.shields.io/badge/os-linux%2C%20win%2C%20mac-pink.svg"></a>
<a href=""><img src="https://img.shields.io/pypi/format/PaddleOCR?color=c77"></a>
<a href="https://github.com/PaddlePaddle/PaddleOCR/graphs/contributors"><img src="https://img.shields.io/github/contributors/PaddlePaddle/PaddleOCR?color=9ea"></a>
<a href="https://pypi.org/project/PaddleOCR/"><img src="https://img.shields.io/pypi/dm/PaddleOCR?color=9cf"></a>
<a href="https://github.com/PaddlePaddle/PaddleOCR/stargazers"><img src="https://img.shields.io/github/stars/PaddlePaddle/PaddleOCR?color=ccf"></a>
</p>
## 简介
PaddleOCR旨在打造一套丰富、领先、且实用的OCR工具库，助力使用者训练出更好的模型，并应用落地。
## 注意
PaddleOCR同时支持动态图与静态图两种编程范式
- 动态图版本：dygraph分支（默认），需将paddle版本升级至2.0.0（[快速安装](./doc/doc_ch/installation.md)）
- 静态图版本：develop分支
**近期更新**
- 2021.4.8 release 2.1版本，新增AAAI 2021论文[端到端识别算法PGNet](./doc/doc_ch/pgnet.md)开源，[多语言模型](./doc/doc_ch/multi_languages.md)支持种类增加到80+。
- 2021.2.1 [FAQ](./doc/doc_ch/FAQ.md)新增5个高频问题，总数162个，每周一都会更新，欢迎大家持续关注。
- 2021.1.21 更新多语言识别模型，目前支持语种超过27种，包括中文简体、中文繁体、英文、法文、德文、韩文、日文、意大利文、西班牙文、葡萄牙文、俄罗斯文、阿拉伯文等，后续计划可以参考[多语言研发计划](https://github.com/PaddlePaddle/PaddleOCR/issues/1048)
- 2020.12.15 更新数据合成工具[Style-Text](./StyleText/README_ch.md)，可以批量合成大量与目标场景类似的图像，在多个场景验证，效果明显提升。
- 2020.11.25 更新半自动标注工具[PPOCRLabel](./PPOCRLabel/README_ch.md)，辅助开发者高效完成标注任务，输出格式与PP-OCR训练任务完美衔接。
- 2020.9.22 更新PP-OCR技术文章，https://arxiv.org/abs/2009.09941
- [More](./doc/doc_ch/update.md)
- PaddleOCR研发团队对最新发版内容技术深入解读，9月8日晚上20:15，[直播地址](https://live.bilibili.com/21689802)。
- 2021.9.7 发布PaddleOCR v2.3，发布[PP-OCRv2](#PP-OCRv2)，CPU推理速度相比于PP-OCR server提升220%；效果相比于PP-OCR mobile 提升7%。
- 2021.8.3 发布PaddleOCR v2.2，新增文档结构分析[PP-Structure](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.2/ppstructure/README_ch.md)工具包，支持版面分析与表格识别（含Excel导出）。
- 2021.6.29 [FAQ](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.2/doc/doc_ch/FAQ.md)新增5个高频问题，总数248个，每周一都会更新，欢迎大家持续关注。
- 2021.4.8 release 2.1版本，新增AAAI 2021论文[端到端识别算法PGNet](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.2/doc/doc_ch/pgnet.md)开源，[多语言模型](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.2/doc/doc_ch/multi_languages.md)支持种类增加到80+。
- [More](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.2/doc/doc_ch/update.md)
## 特性
- PPOCR系列高质量预训练模型，准确的识别效果
- 超轻量ppocr_mobile移动端系列：检测（3.0M）+方向分类器（1.4M）+ 识别（5.0M）= 9.4M
- 通用ppocr_server系列：检测（47.1M）+方向分类器（1.4M）+ 识别（94.9M）= 143.4M
- PP-OCR系列高质量预训练模型，准确的识别效果
- 超轻量PP-OCRv2系列：检测（3.1M）+ 方向分类器（1.4M）+ 识别（8.5M）= 13.0M
- 超轻量PP-OCR mobile移动端系列：检测（3.0M）+方向分类器（1.4M）+ 识别（5.0M）= 9.4M
- 通用PPOCR server系列：检测（47.1M）+方向分类器（1.4M）+ 识别（94.9M）= 143.4M
- 支持中英文数字组合识别、竖排文本识别、长文本识别
- 支持多语言识别：韩语、日语、德语、法语
- 丰富易用的OCR相关工具组件
- 半自动数据标注工具PPOCRLabel：支持快速高效的数据标注
- 数据合成工具Style-Text：批量合成大量与目标场景类似的图像
- 文档分析能力PP-Structure：版面分析与表格识别
- 支持用户自定义训练，提供丰富的预测推理部署方案
- 支持PIP快速安装使用
- 可运行于Linux、Windows、MacOS等多种系统
......@@ -39,7 +54,7 @@ PaddleOCR同时支持动态图与静态图两种编程范式
<img src="doc/imgs_results/ch_ppocr_mobile_v2.0/00018069.jpg" width="800">
</div>
上图是通用ppocr_server模型效果展示，更多效果图请见[效果展示页面](./doc/doc_ch/visualization.md)。
上图是通用PP-OCR server模型效果展示，更多效果图请见[效果展示页面](./doc/doc_ch/visualization.md)。
<a name="欢迎加入PaddleOCR技术交流群"></a>
## 欢迎加入PaddleOCR技术交流群
......@@ -62,71 +77,78 @@ PaddleOCR同时支持动态图与静态图两种编程范式
- 代码体验：从[快速安装](./doc/doc_ch/quickstart.md) 开始
<a name="模型下载"></a>
## PP-OCR 2.0系列模型列表（更新中）
**说明** ：2.0版模型和[1.1版模型](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/doc/doc_ch/models_list.md)的主要区别在于动态图训练vs.静态图训练，模型性能上无明显差距。
## PP-OCR系列模型列表（更新中）
| 模型简介 | 模型名称 |推荐场景 | 检测模型 | 方向分类器 | 识别模型 |
| ------------ | --------------- | ----------------|---- | ---------- | -------- |
| 中英文超轻量OCR模型（9.4M） | ch_ppocr_mobile_v2.0_xx |移动端&服务器端|[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_det_infer.tar) / [预训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_det_train.tar)|[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_infer.tar) / [预训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_train.tar) |[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_infer.tar) / [预训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_pre.tar) |
| 中英文通用OCR模型（143.4M） |ch_ppocr_server_v2.0_xx|服务器端 |[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_det_infer.tar) / [预训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_det_train.tar) |[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_infer.tar) / [预训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_train.tar) |[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_rec_infer.tar) / [预训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_rec_pre.tar) |
| 中英文超轻量PP-OCRv2模型（13.0M） | ch_PP-OCRv2_xx |移动端&服务器端|[推理模型](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_det_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/chinese/ch_PP-OCRv2_det_distill_train.tar)| [推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_infer.tar) / [预训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_train.tar) |[推理模型](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_rec_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_rec_train.tar)|
| 中英文超轻量PP-OCR mobile模型（9.4M） | ch_ppocr_mobile_v2.0_xx |移动端&服务器端|[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_det_infer.tar) / [预训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_det_train.tar)|[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_infer.tar) / [预训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_train.tar) |[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_infer.tar) / [预训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_pre.tar) |
| 中英文通用PP-OCR server模型（143.4M） |ch_ppocr_server_v2.0_xx|服务器端 |[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_det_infer.tar) / [预训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_det_train.tar) |[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_infer.tar) / [预训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_train.tar) |[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_rec_infer.tar) / [预训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_rec_pre.tar) |
更多模型下载（包括多语言），可以参考[PP-OCR v2.0 系列模型下载](./doc/doc_ch/models_list.md)
更多模型下载（包括多语言），可以参考[PP-OCR 系列模型下载](./doc/doc_ch/models_list.md)
## 文档教程
- [快速安装](./doc/doc_ch/installation.md)
- [中文OCR模型快速使用](./doc/doc_ch/quickstart.md)
- [多语言OCR模型快速使用](./doc/doc_ch/multi_languages.md)
- [代码组织结构](./doc/doc_ch/tree.md)
- 算法介绍
- [文本检测](./doc/doc_ch/algorithm_overview.md)
- [文本识别](./doc/doc_ch/algorithm_overview.md)
- [PP-OCR Pipeline](#PP-OCR)
- [运行环境准备](./doc/doc_ch/environment.md)
- [快速开始（中英文/多语言/文档分析）](./doc/doc_ch/quickstart.md)
- [PaddleOCR全景图与项目克隆](./doc/doc_ch/paddleOCR_overview.md)
- PP-OCR产业落地：从训练到部署
- [PP-OCR模型与配置文件](./doc/doc_ch/models_and_config.md)
- [PP-OCR模型下载](./doc/doc_ch/models_list.md)
- [配置文件内容与生成](./doc/doc_ch/config.md)
- [PP-OCR模型库快速推理](./doc/doc_ch/inference_ppocr.md)
- [PP-OCR模型训练](./doc/doc_ch/training.md)
- [文本检测](./doc/doc_ch/detection.md)
- [文本识别](./doc/doc_ch/recognition.md)
- [方向分类器](./doc/doc_ch/angle_class.md)
- PP-OCR模型推理部署
- [基于C++预测引擎推理](./deploy/cpp_infer/readme.md)
- [服务化部署](./deploy/pdserving/README_CN.md)
- [端侧部署](./deploy/lite/readme.md)
- [Benchmark](./doc/doc_ch/benchmark.md)
- [PP-Structure信息提取](./ppstructure/README_ch.md)
- [版面分析](./ppstructure/layout/README_ch.md)
- [表格识别](./ppstructure/table/README_ch.md)
- 数据标注与合成
- [半自动标注工具PPOCRLabel](./PPOCRLabel/README_ch.md)
- [数据合成工具Style-Text](./StyleText/README_ch.md)
- [其它数据标注工具](./doc/doc_ch/data_annotation.md)
- [其它数据合成工具](./doc/doc_ch/data_synthesis.md)
- OCR学术圈
- [两阶段模型介绍与下载](./doc/doc_ch/algorithm_overview.md)
- [端到端PGNet算法](./doc/doc_ch/pgnet.md)
- 模型训练/评估
- [文本检测](./doc/doc_ch/detection.md)
- [文本识别](./doc/doc_ch/recognition.md)
- [方向分类器](./doc/doc_ch/angle_class.md)
- [yml参数配置文件介绍](./doc/doc_ch/config.md)
- 预测部署
- [基于pip安装whl包快速推理](./doc/doc_ch/whl.md)
- [基于Python脚本预测引擎推理](./doc/doc_ch/inference.md)
- [基于C++预测引擎推理](./deploy/cpp_infer/readme.md)
- [服务化部署](./deploy/pdserving/README_CN.md)
- [端侧部署](./deploy/lite/readme.md)
- [Benchmark](./doc/doc_ch/benchmark.md)
- 数据集
- [通用中英文OCR数据集](./doc/doc_ch/datasets.md)
- [手写中文OCR数据集](./doc/doc_ch/handwritten_datasets.md)
- [垂类多语言OCR数据集](./doc/doc_ch/vertical_and_multilingual_datasets.md)
- 数据标注与合成
- [半自动标注工具PPOCRLabel](./PPOCRLabel/README_ch.md)
- [数据合成工具Style-Text](./StyleText/README_ch.md)
- [其它数据标注工具](./doc/doc_ch/data_annotation.md)
- [其它数据合成工具](./doc/doc_ch/data_synthesis.md)
- [效果展示](#效果展示)
- FAQ
- [【精选】OCR精选10个问题](./doc/doc_ch/FAQ.md)
- [【理论篇】OCR通用32个问题](./doc/doc_ch/FAQ.md)
- [【实战篇】PaddleOCR实战110个问题](./doc/doc_ch/FAQ.md)
- [【理论篇】OCR通用50个问题](./doc/doc_ch/FAQ.md)
- [【实战篇】PaddleOCR实战183个问题](./doc/doc_ch/FAQ.md)
- [技术交流群](#欢迎加入PaddleOCR技术交流群)
- [参考文献](./doc/doc_ch/reference.md)
- [许可证书](#许可证书)
- [贡献代码](#贡献代码)
- [代码组织结构](./doc/doc_ch/tree.md)
<a name="PP-OCRv2"></a>
<a name="PP-OCR"></a>
## PP-OCR Pipeline
## PP-OCRv2 Pipeline
<div align="center">
<img src="./doc/ppocr_framework.png" width="800">
<img src="./doc/ppocrv2_framework.jpg" width="800">
</div>
PP-OCR是一个实用的超轻量OCR系统。主要由DB文本检测[2]、检测框矫正和CRNN文本识别三部分组成[7]。该系统从骨干网络选择和调整、预测头部的设计、数据增强、学习率变换策略、正则化参数选择、预训练模型使用以及模型自动裁剪量化8个方面，采用19个有效策略，对各个模块的模型进行效果调优和瘦身，最终得到整体大小为3.5M的超轻量中英文OCR和2.8M的英文数字OCR。更多细节请参考PP-OCR技术方案 https://arxiv.org/abs/2009.09941 。其中FPGM裁剪器[8]和PACT量化[9]的实现可以参考[PaddleSlim](https://github.com/PaddlePaddle/PaddleSlim)。
[1] PP-OCR是一个实用的超轻量OCR系统。主要由DB文本检测、检测框矫正和CRNN文本识别三部分组成。该系统从骨干网络选择和调整、预测头部的设计、数据增强、学习率变换策略、正则化参数选择、预训练模型使用以及模型自动裁剪量化8个方面，采用19个有效策略，对各个模块的模型进行效果调优和瘦身(如绿框所示)，最终得到整体大小为3.5M的超轻量中英文OCR和2.8M的英文数字OCR。更多细节请参考PP-OCR技术方案 https://arxiv.org/abs/2009.09941
[2] PP-OCRv2在PP-OCR的基础上，进一步在5个方面重点优化，检测模型采用CML协同互学习知识蒸馏策略和CopyPaste数据增广策略；识别模型采用LCNet轻量级骨干网络、UDML 改进知识蒸馏策略和Enhanced CTC loss损失函数改进（如上图红框所示），进一步在推理速度和预测效果上取得明显提升。更多细节请参考PP-OCR技术方案（arxiv链接生成中）。
<a name="效果展示"></a>
## 效果展示 [more](./doc/doc_ch/visualization.md)
- 中文模型
<div align="center">
<img src="./doc/imgs_results/ch_ppocr_mobile_v2.0/test_add_91.jpg" width="800">
<img src="./doc/imgs_results/ch_ppocr_mobile_v2.0/00015504.jpg" width="800">
<img src="./doc/imgs_results/ch_ppocr_mobile_v2.0/00056221.jpg" width="800">
<img src="./doc/imgs_results/ch_ppocr_mobile_v2.0/rotate_00052204.jpg" width="800">
</div>
......
configs/det/ch_ppocr_v2.1/ch_det_lite_train_cml_v2.1.yml → configs/det/ch_PP-OCRv2/ch_PP-OCR_det_cml.yml
浏览文件 @ a8a9b2e5
......@@ -8,7 +8,7 @@ Global:
# evaluation is run every 5000 iterations after the 4000th iteration
eval_batch_step: [3000, 2000]
cal_metric_during_train: False
pretrained_model: ./pretrain_models/MobileNetV3_large_x0_5_pretrained
pretrained_model: ./pretrain_models/ch_PP-OCRv2_det_distill_train/best_accuracy
checkpoints:
save_inference_dir:
use_visualdl: False
......@@ -19,30 +19,26 @@ Architecture:
name: DistillationModel
algorithm: Distillation
Models:
Student:
pretrained: ./pretrain_models/MobileNetV3_large_x0_5_pretrained
freeze_params: false
Teacher:
freeze_params: true
return_all_feats: false
model_type: det
algorithm: DB
Transform:
Backbone:
name: MobileNetV3
scale: 0.5
model_name: large
disable_se: True
name: ResNet
layers: 18
Neck:
name: DBFPN
out_channels: 96
out_channels: 256
Head:
name: DBHead
k: 50
Student2:
pretrained: ./pretrain_models/MobileNetV3_large_x0_5_pretrained
Student:
freeze_params: false
return_all_feats: false
model_type: det
algorithm: DB
Transform:
Backbone:
name: MobileNetV3
scale: 0.5
......@@ -54,23 +50,24 @@ Architecture:
Head:
name: DBHead
k: 50
Teacher:
pretrained: ./pretrain_models/ch_ppocr_server_v2.0_det_train/best_accuracy
freeze_params: true
Student2:
freeze_params: false
return_all_feats: false
model_type: det
algorithm: DB
Transform:
Backbone:
name: ResNet
layers: 18
name: MobileNetV3
scale: 0.5
model_name: large
disable_se: True
Neck:
name: DBFPN
out_channels: 256
out_channels: 96
Head:
name: DBHead
k: 50
Loss:
name: CombinedLoss
loss_config_list:
......
configs/det/ch_PP-OCRv2/ch_PP-OCR_det_student.yml 0 → 100644
浏览文件 @ a8a9b2e5
Global:
use_gpu: true
epoch_num: 1200
log_smooth_window: 20
print_batch_step: 10
save_model_dir: ./output/ch_db_mv3/
save_epoch_step: 1200
# evaluation is run every 5000 iterations after the 4000th iteration
eval_batch_step: [0, 400]
cal_metric_during_train: False
pretrained_model: ./pretrain_models/student.pdparams
checkpoints:
save_inference_dir:
use_visualdl: False
infer_img: doc/imgs_en/img_10.jpg
save_res_path: ./output/det_db/predicts_db.txt
Architecture:
model_type: det
algorithm: DB
Transform:
Backbone:
name: MobileNetV3
scale: 0.5
model_name: large
disable_se: True
Neck:
name: DBFPN
out_channels: 96
Head:
name: DBHead
k: 50
Loss:
name: DBLoss
balance_loss: true
main_loss_type: DiceLoss
alpha: 5
beta: 10
ohem_ratio: 3
Optimizer:
name: Adam
beta1: 0.9
beta2: 0.999
lr:
name: Cosine
learning_rate: 0.001
warmup_epoch: 2
regularizer:
name: 'L2'
factor: 0
PostProcess:
name: DBPostProcess
thresh: 0.3
box_thresh: 0.6
max_candidates: 1000
unclip_ratio: 1.5
Metric:
name: DetMetric
main_indicator: hmean
Train:
dataset:
name: SimpleDataSet
data_dir: ./train_data/icdar2015/text_localization/
label_file_list:
- ./train_data/icdar2015/text_localization/train_icdar2015_label.txt
ratio_list: [1.0]
transforms:
- DecodeImage: # load image
img_mode: BGR
channel_first: False
- DetLabelEncode: # Class handling label
- IaaAugment:
augmenter_args:
- { 'type': Fliplr, 'args': { 'p': 0.5 } }
- { 'type': Affine, 'args': { 'rotate': [-10, 10] } }
- { 'type': Resize, 'args': { 'size': [0.5, 3] } }
- EastRandomCropData:
size: [960, 960]
max_tries: 50
keep_ratio: true
- MakeBorderMap:
shrink_ratio: 0.4
thresh_min: 0.3
thresh_max: 0.7
- MakeShrinkMap:
shrink_ratio: 0.4
min_text_size: 8
- NormalizeImage:
scale: 1./255.
mean: [0.485, 0.456, 0.406]
std: [0.229, 0.224, 0.225]
order: 'hwc'
- ToCHWImage:
- KeepKeys:
keep_keys: ['image', 'threshold_map', 'threshold_mask', 'shrink_map', 'shrink_mask'] # the order of the dataloader list
loader:
shuffle: True
drop_last: False
batch_size_per_card: 8
num_workers: 4
Eval:
dataset:
name: SimpleDataSet
data_dir: ./train_data/icdar2015/text_localization/
label_file_list:
- ./train_data/icdar2015/text_localization/test_icdar2015_label.txt
transforms:
- DecodeImage: # load image
img_mode: BGR
channel_first: False
- DetLabelEncode: # Class handling label
- DetResizeForTest:
# image_shape: [736, 1280]
- NormalizeImage:
scale: 1./255.
mean: [0.485, 0.456, 0.406]
std: [0.229, 0.224, 0.225]
order: 'hwc'
- ToCHWImage:
- KeepKeys:
keep_keys: ['image', 'shape', 'polys', 'ignore_tags']
loader:
shuffle: False
drop_last: False
batch_size_per_card: 1 # must be 1
num_workers: 2
configs/det/det_mv3_db.yml
浏览文件 @ a8a9b2e5
......@@ -128,4 +128,4 @@ Eval:
drop_last: False
batch_size_per_card: 1 # must be 1
num_workers: 8
use_shared_memory: False
\ No newline at end of file
use_shared_memory: False
configs/det/det_r50_vd_db.yml
浏览文件 @ a8a9b2e5
......@@ -98,7 +98,7 @@ Train:
shuffle: True
drop_last: False
batch_size_per_card: 16
num_workers: 8
num_workers: 4
Eval:
dataset:
......@@ -125,4 +125,4 @@ Eval:
shuffle: False
drop_last: False
batch_size_per_card: 1 # must be 1
num_workers: 8
\ No newline at end of file
num_workers: 8
configs/rec/rec_icdar15_train.yml
浏览文件 @ a8a9b2e5
......@@ -14,8 +14,8 @@ Global:
use_visualdl: False
infer_img: doc/imgs_words_en/word_10.png
# for data or label process
character_dict_path: ppocr/utils/ic15_dict.txt
character_type: ch
character_dict_path: ppocr/utils/en_dict.txt
character_type: EN
max_text_length: 25
infer_mode: False
use_space_char: False
......
configs/rec/rec_mtb_nrtr.yml 0 → 100644
浏览文件 @ a8a9b2e5
Global:
use_gpu: True
epoch_num: 21
log_smooth_window: 20
print_batch_step: 10
save_model_dir: ./output/rec/nrtr/
save_epoch_step: 1
# evaluation is run every 2000 iterations
eval_batch_step: [0, 2000]
cal_metric_during_train: True
pretrained_model:
checkpoints:
save_inference_dir:
use_visualdl: False
infer_img: doc/imgs_words_en/word_10.png
# for data or label process
character_dict_path:
character_type: EN_symbol
max_text_length: 25
infer_mode: False
use_space_char: True
save_res_path: ./output/rec/predicts_nrtr.txt
Optimizer:
name: Adam
beta1: 0.9
beta2: 0.99
clip_norm: 5.0
lr:
name: Cosine
learning_rate: 0.0005
warmup_epoch: 2
regularizer:
name: 'L2'
factor: 0.
Architecture:
model_type: rec
algorithm: NRTR
in_channels: 1
Transform:
Backbone:
name: MTB
cnn_num: 2
Head:
name: Transformer
d_model: 512
num_encoder_layers: 6
beam_size: 10 # When Beam size is greater than 0, it means to use beam search when evaluation.
Loss:
name: NRTRLoss
smoothing: True
PostProcess:
name: NRTRLabelDecode
Metric:
name: RecMetric
main_indicator: acc
Train:
dataset:
name: LMDBDataSet
data_dir: ./train_data/data_lmdb_release/training/
transforms:
- NRTRDecodeImage: # load image
img_mode: BGR
channel_first: False
- NRTRLabelEncode: # Class handling label
- NRTRRecResizeImg:
image_shape: [100, 32]
resize_type: PIL # PIL or OpenCV
- KeepKeys:
keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order
loader:
shuffle: True
batch_size_per_card: 512
drop_last: True
num_workers: 8
Eval:
dataset:
name: LMDBDataSet
data_dir: ./train_data/data_lmdb_release/evaluation/
transforms:
- NRTRDecodeImage: # load image
img_mode: BGR
channel_first: False
- NRTRLabelEncode: # Class handling label
- NRTRRecResizeImg:
image_shape: [100, 32]
resize_type: PIL # PIL or OpenCV
- KeepKeys:
keep_keys: ['image', 'label', 'length'] # dataloader will return list in this order
loader:
shuffle: False
drop_last: False
batch_size_per_card: 256
num_workers: 1
use_shared_memory: False
deploy/cpp_infer/CMakeLists.txt
浏览文件 @ a8a9b2e5
project(ppocr CXX C)
cmake_minimum_required(VERSION 3.14)
option(WITH_MKL "Compile demo with MKL/OpenBlas support, default use MKL." ON)
option(WITH_GPU "Compile demo with GPU/CPU, default use CPU." OFF)
......@@ -206,9 +207,12 @@ endif()
set(DEPS ${DEPS} ${OpenCV_LIBS})
include(FetchContent)
include(external-cmake/auto-log.cmake)
include_directories(${FETCHCONTENT_BASE_DIR}/extern_autolog-src)
AUX_SOURCE_DIRECTORY(./src SRCS)
add_executable(${DEMO_NAME} ${SRCS})
target_link_libraries(${DEMO_NAME} ${DEPS})
if (WIN32 AND WITH_MKL)
......
deploy/cpp_infer/docs/vs2019_build_withgpu_config.png 0 → 100644
浏览文件 @ a8a9b2e5
因为 它太大了无法显示 image diff 。你可以改为 查看blob。
deploy/cpp_infer/docs/windows_vs2019_build.md
浏览文件 @ a8a9b2e5
......@@ -5,20 +5,20 @@ PaddleOCR在Windows 平台下基于`Visual Studio 2019 Community` 进行了测
## 前置条件
* Visual Studio 2019
* CUDA 9.0 / CUDA 10.0，cudnn 7+ （仅在使用GPU版本的预测库时需要）
* CUDA 10.2，cudnn 7+ （仅在使用GPU版本的预测库时需要）
* CMake 3.0+
请确保系统已经安装好上述基本软件，我们使用的是`VS2019`的社区版。
**下面所有示例以工作目录为 `D:\projects`演示**。
### Step1: 下载PaddlePaddle C++ 预测库 fluid_inference
### Step1: 下载PaddlePaddle C++ 预测库 paddle_inference
PaddlePaddle C++ 预测库针对不同的`CPU`和`CUDA`版本提供了不同的预编译版本，请根据实际情况下载: [C++预测库下载列表](https://paddleinference.paddlepaddle.org.cn/user_guides/download_lib.html#windows)
解压后`D:\projects\fluid_inference`目录包含内容为：
解压后`D:\projects\paddle_inference`目录包含内容为：
```
fluid_inference
paddle_inference
├── paddle # paddle核心库和头文件
|
├── third_party # 第三方依赖库和头文件
......@@ -46,13 +46,13 @@ fluid_inference
![step2.2](https://paddleseg.bj.bcebos.com/inference/vs2019_step3.png)
3. 点击：`项目`->`cpp_inference_demo的CMake设置`
3. 点击：`项目`->`CMake设置`
![step3](https://paddleseg.bj.bcebos.com/inference/vs2019_step4.png)
4. 点击`浏览`，分别设置编译选项指定`CUDA`、`CUDNN_LIB`、`OpenCV`、`Paddle预测库`的路径
4. 分别设置编译选项指定`CUDA`、`CUDNN_LIB`、`OpenCV`、`Paddle预测库`的路径
三个编译参数的含义说明如下（带`*`表示仅在使用**GPU版本**预测库时指定, 其中CUDA库版本尽量对齐，**使用9.0、10.0版本，不使用9.2、10.1等版本CUDA库**）：
三个编译参数的含义说明如下（带`*`表示仅在使用**GPU版本**预测库时指定, 其中CUDA库版本尽量对齐）：
| 参数名 | 含义 |
| ---- | ---- |
......@@ -67,6 +67,11 @@ fluid_inference
![step4](https://paddleseg.bj.bcebos.com/inference/vs2019_step5.png)
下面给出with GPU的配置示例：
![step5](./vs2019_build_withgpu_config.png)
**注意：**
CMAKE_BACKWARDS的版本要根据平台安装cmake的版本进行设置。
**设置完成后**, 点击上图中`保存并生成CMake缓存以加载变量`。
5. 点击`生成`->`全部生成`
......@@ -74,24 +79,34 @@ fluid_inference
![step6](https://paddleseg.bj.bcebos.com/inference/vs2019_step6.png)
### Step4: 预测及可视化
### Step4: 预测
上述`Visual Studio 2019`编译产出的可执行文件在`out\build\x64-Release`目录下，打开`cmd`，并切换到该目录：
上述`Visual Studio 2019`编译产出的可执行文件在`out\build\x64-Release\Release`目录下，打开`cmd`，并切换到`D:\projects\PaddleOCR\deploy\cpp_infer\`：
```
cd D:\projects\PaddleOCR\deploy\cpp_infer\out\build\x64-Release
cd D:\projects\PaddleOCR\deploy\cpp_infer
```
可执行文件`ocr_system.exe`即为样例的预测程序，其主要使用方法如下
可执行文件`ppocr.exe`即为样例的预测程序，其主要使用方法如下，更多使用方法可以参考[说明文档](../readme.md)`运行demo`部分。
```shell
#预测图片 `D:\projects\PaddleOCR\doc\imgs\10.jpg`
.\ocr_system.exe D:\projects\PaddleOCR\deploy\cpp_infer\tools\config.txt D:\projects\PaddleOCR\doc\imgs\10.jpg
#识别中文图片 `D:\projects\PaddleOCR\doc\imgs_words\ch\`
.\out\build\x64-Release\Release\ppocr.exe rec --rec_model_dir=D:\projects\PaddleOCR\ch_ppocr_mobile_v2.0_rec_infer --image_dir=D:\projects\PaddleOCR\doc\imgs_words\ch\
#识别英文图片 'D:\projects\PaddleOCR\doc\imgs_words\en\'
.\out\build\x64-Release\Release\ppocr.exe rec --rec_model_dir=D:\projects\PaddleOCR\inference\rec_mv3crnn --image_dir=D:\projects\PaddleOCR\doc\imgs_words\en\ --char_list_file=D:\projects\PaddleOCR\ppocr\utils\dict\en_dict.txt
```
第一个参数为配置文件路径，第二个参数为需要预测的图片路径。
第一个参数为配置文件路径，第二个参数为需要预测的图片路径，第三个参数为配置文本识别的字典。
### 注意
### FQA
* 在Windows下的终端中执行文件exe时，可能会发生乱码的现象，此时需要在终端中输入`CHCP 65001`，将终端的编码方式由GBK编码(默认)改为UTF-8编码，更加具体的解释可以参考这篇博客：[https://blog.csdn.net/qq_35038153/article/details/78430359](https://blog.csdn.net/qq_35038153/article/details/78430359)。
* 编译时，如果报错`错误：C1083 无法打开包括文件:"dirent.h":No such file or directory`，可以参考该[文档](https://blog.csdn.net/Dora_blank/article/details/117740837#41_C1083_direnthNo_such_file_or_directory_54)，新建`dirent.h`文件，并添加到`VC++`的包含目录中。
* 编译时，如果报错`错误：C1083 无法打开包括文件:"dirent.h":No such file or directory`，可以参考该[文档](https://blog.csdn.net/Dora_blank/article/details/117740837#41_C1083_direnthNo_such_file_or_directory_54)，新建`dirent.h`文件，并添加到`utility.cpp`的头文件引用中。同时修改`utility.cpp`70行：`lstat`改成`stat`。
* 编译时，如果报错`Autolog未定义`，新建`autolog.h`文件，内容为：[autolog.h](https://github.com/LDOUBLEV/AutoLog/blob/main/auto_log/autolog.h)，并添加到`main.cpp`的头文件引用中，再次编译。
* 运行时，如果弹窗报错找不到`paddle_inference.dll`或者`openblas.dll`，在`D:\projects\paddle_inference`预测库内找到这两个文件，复制到`D:\projects\PaddleOCR\deploy\cpp_infer\out\build\x64-Release\Release`目录下。不用重新编译，再次运行即可。
* 运行时，弹窗报错提示`应用程序无法正常启动(0xc0000142)`，并且`cmd`窗口内提示`You are using Paddle compiled with TensorRT, but TensorRT dynamic library is not found.`，把tensort目录下的lib里面的所有dll文件复制到release目录下，再次运行即可。
deploy/cpp_infer/external-cmake/auto-log.cmake 0 → 100644
浏览文件 @ a8a9b2e5
find_package(Git REQUIRED)
include(FetchContent)
set(FETCHCONTENT_BASE_DIR "${CMAKE_CURRENT_BINARY_DIR}/third-party")
FetchContent_Declare(
extern_Autolog
PREFIX autolog
GIT_REPOSITORY https://github.com/LDOUBLEV/AutoLog.git
GIT_TAG main
)
FetchContent_MakeAvailable(extern_Autolog)
deploy/cpp_infer/include/ocr_cls.h
浏览文件 @ a8a9b2e5
......@@ -42,7 +42,7 @@ public:
const int &gpu_id, const int &gpu_mem,
const int &cpu_math_library_num_threads,
const bool &use_mkldnn, const double &cls_thresh,
const bool &use_tensorrt, const bool &use_fp16) {
const bool &use_tensorrt, const std::string &precision) {
this->use_gpu_ = use_gpu;
this->gpu_id_ = gpu_id;
this->gpu_mem_ = gpu_mem;
......@@ -51,7 +51,7 @@ public:
this->cls_thresh = cls_thresh;
this->use_tensorrt_ = use_tensorrt;
this->use_fp16_ = use_fp16;
this->precision_ = precision;
LoadModel(model_dir);
}
......@@ -75,7 +75,7 @@ private:
std::vector<float> scale_ = {1 / 0.5f, 1 / 0.5f, 1 / 0.5f};
bool is_scale_ = true;
bool use_tensorrt_ = false;
bool use_fp16_ = false;
std::string precision_ = "fp32";
// pre-process
ClsResizeImg resize_op_;
Normalize normalize_op_;
......
deploy/cpp_infer/include/ocr_det.h
浏览文件 @ a8a9b2e5
......@@ -46,7 +46,7 @@ public:
const double &det_db_box_thresh,
const double &det_db_unclip_ratio,
const bool &use_polygon_score, const bool &visualize,
const bool &use_tensorrt, const bool &use_fp16) {
const bool &use_tensorrt, const std::string &precision) {
this->use_gpu_ = use_gpu;
this->gpu_id_ = gpu_id;
this->gpu_mem_ = gpu_mem;
......@@ -62,7 +62,7 @@ public:
this->visualize_ = visualize;
this->use_tensorrt_ = use_tensorrt;
this->use_fp16_ = use_fp16;
this->precision_ = precision;
LoadModel(model_dir);
}
......@@ -71,7 +71,7 @@ public:
void LoadModel(const std::string &model_dir);
// Run predictor
void Run(cv::Mat &img, std::vector<std::vector<std::vector<int>>> &boxes);
void Run(cv::Mat &img, std::vector<std::vector<std::vector<int>>> &boxes, std::vector<double> *times);
private:
std::shared_ptr<Predictor> predictor_;
......@@ -91,7 +91,7 @@ private:
bool visualize_ = true;
bool use_tensorrt_ = false;
bool use_fp16_ = false;
std::string precision_ = "fp32";
std::vector<float> mean_ = {0.485f, 0.456f, 0.406f};
std::vector<float> scale_ = {1 / 0.229f, 1 / 0.224f, 1 / 0.225f};
......
deploy/cpp_infer/include/ocr_rec.h
浏览文件 @ a8a9b2e5
......@@ -44,14 +44,14 @@ public:
const int &gpu_id, const int &gpu_mem,
const int &cpu_math_library_num_threads,
const bool &use_mkldnn, const string &label_path,
const bool &use_tensorrt, const bool &use_fp16) {
const bool &use_tensorrt, const std::string &precision) {
this->use_gpu_ = use_gpu;
this->gpu_id_ = gpu_id;
this->gpu_mem_ = gpu_mem;
this->cpu_math_library_num_threads_ = cpu_math_library_num_threads;
this->use_mkldnn_ = use_mkldnn;
this->use_tensorrt_ = use_tensorrt;
this->use_fp16_ = use_fp16;
this->precision_ = precision;
this->label_list_ = Utility::ReadDict(label_path);
this->label_list_.insert(this->label_list_.begin(),
......@@ -64,7 +64,7 @@ public:
// Load Paddle inference model
void LoadModel(const std::string &model_dir);
void Run(cv::Mat &img);
void Run(cv::Mat &img, std::vector<double> *times);
private:
std::shared_ptr<Predictor> predictor_;
......@@ -81,7 +81,7 @@ private:
std::vector<float> scale_ = {1 / 0.5f, 1 / 0.5f, 1 / 0.5f};
bool is_scale_ = true;
bool use_tensorrt_ = false;
bool use_fp16_ = false;
std::string precision_ = "fp32";
// pre-process
CrnnResizeImg resize_op_;
Normalize normalize_op_;
......@@ -90,9 +90,6 @@ private:
// post-process
PostProcessor post_processor_;
cv::Mat GetRotateCropImage(const cv::Mat &srcimage,
std::vector<std::vector<int>> box);
}; // class CrnnRecognizer
} // namespace PaddleOCR
deploy/cpp_infer/include/utility.h
浏览文件 @ a8a9b2e5
......@@ -47,6 +47,9 @@ public:
static void GetAllFiles(const char *dir_name,
std::vector<std::string> &all_inputs);
static cv::Mat GetRotateCropImage(const cv::Mat &srcimage,
std::vector<std::vector<int>> box);
};
} // namespace PaddleOCR
\ No newline at end of file
deploy/cpp_infer/src/main.cpp
浏览文件 @ a8a9b2e5
......@@ -31,17 +31,21 @@
#include <include/ocr_det.h>
#include <include/ocr_cls.h>
#include <include/ocr_rec.h>
#include <include/utility.h>
#include <sys/stat.h>
#include <gflags/gflags.h>
#include "auto_log/autolog.h"
DEFINE_bool(use_gpu, false, "Infering with GPU or CPU.");
DEFINE_int32(gpu_id, 0, "Device id of GPU to execute.");
DEFINE_int32(gpu_mem, 4000, "GPU id when infering with GPU.");
DEFINE_int32(cpu_math_library_num_threads, 10, "Num of threads with CPU.");
DEFINE_bool(use_mkldnn, false, "Whether use mkldnn with CPU.");
DEFINE_int32(cpu_threads, 10, "Num of threads with CPU.");
DEFINE_bool(enable_mkldnn, false, "Whether use mkldnn with CPU.");
DEFINE_bool(use_tensorrt, false, "Whether use tensorrt.");
DEFINE_bool(use_fp16, false, "Whether use fp16 when use tensorrt.");
DEFINE_string(precision, "fp32", "Precision be one of fp32/fp16/int8");
DEFINE_bool(benchmark, true, "Whether use benchmark.");
DEFINE_string(save_log_path, "./log_output/", "Save benchmark log path.");
// detection related
DEFINE_string(image_dir, "", "Dir of input image.");
DEFINE_string(det_model_dir, "", "Path of det inference model.");
......@@ -57,6 +61,7 @@ DEFINE_string(cls_model_dir, "", "Path of cls inference model.");
DEFINE_double(cls_thresh, 0.9, "Threshold of cls_thresh.");
// recognition related
DEFINE_string(rec_model_dir, "", "Path of rec inference model.");
DEFINE_int32(rec_batch_num, 1, "rec_batch_num.");
DEFINE_string(char_list_file, "../../ppocr/utils/ppocr_keys_v1.txt", "Path of dictionary.");
......@@ -76,88 +81,15 @@ static bool PathExists(const std::string& path){
}
cv::Mat GetRotateCropImage(const cv::Mat &srcimage,
std::vector<std::vector<int>> box) {
cv::Mat image;
srcimage.copyTo(image);
std::vector<std::vector<int>> points = box;
int x_collect[4] = {box[0][0], box[1][0], box[2][0], box[3][0]};
int y_collect[4] = {box[0][1], box[1][1], box[2][1], box[3][1]};
int left = int(*std::min_element(x_collect, x_collect + 4));
int right = int(*std::max_element(x_collect, x_collect + 4));
int top = int(*std::min_element(y_collect, y_collect + 4));
int bottom = int(*std::max_element(y_collect, y_collect + 4));
cv::Mat img_crop;
image(cv::Rect(left, top, right - left, bottom - top)).copyTo(img_crop);
for (int i = 0; i < points.size(); i++) {
points[i][0] -= left;
points[i][1] -= top;
}
int img_crop_width = int(sqrt(pow(points[0][0] - points[1][0], 2) +
pow(points[0][1] - points[1][1], 2)));
int img_crop_height = int(sqrt(pow(points[0][0] - points[3][0], 2) +
pow(points[0][1] - points[3][1], 2)));
cv::Point2f pts_std[4];
pts_std[0] = cv::Point2f(0., 0.);
pts_std[1] = cv::Point2f(img_crop_width, 0.);
pts_std[2] = cv::Point2f(img_crop_width, img_crop_height);
pts_std[3] = cv::Point2f(0.f, img_crop_height);
cv::Point2f pointsf[4];
pointsf[0] = cv::Point2f(points[0][0], points[0][1]);
pointsf[1] = cv::Point2f(points[1][0], points[1][1]);
pointsf[2] = cv::Point2f(points[2][0], points[2][1]);
pointsf[3] = cv::Point2f(points[3][0], points[3][1]);
cv::Mat M = cv::getPerspectiveTransform(pointsf, pts_std);
cv::Mat dst_img;
cv::warpPerspective(img_crop, dst_img, M,
cv::Size(img_crop_width, img_crop_height),
cv::BORDER_REPLICATE);
if (float(dst_img.rows) >= float(dst_img.cols) * 1.5) {
cv::Mat srcCopy = cv::Mat(dst_img.rows, dst_img.cols, dst_img.depth());
cv::transpose(dst_img, srcCopy);
cv::flip(srcCopy, srcCopy, 0);
return srcCopy;
} else {
return dst_img;
}
}
int main_det(int argc, char **argv) {
// Parsing command-line
google::ParseCommandLineFlags(&argc, &argv, true);
if (FLAGS_det_model_dir.empty() || FLAGS_image_dir.empty()) {
std::cout << "Usage[det]: ./ppocr --det_model_dir=/PATH/TO/DET_INFERENCE_MODEL/ "
<< "--image_dir=/PATH/TO/INPUT/IMAGE/" << std::endl;
exit(1);
}
if (!PathExists(FLAGS_image_dir)) {
std::cerr << "[ERROR] image path not exist! image_dir: " << FLAGS_image_dir << endl;
exit(1);
}
std::vector<cv::String> cv_all_img_names;
cv::glob(FLAGS_image_dir, cv_all_img_names);
std::cout << "total images num: " << cv_all_img_names.size() << endl;
int main_det(std::vector<cv::String> cv_all_img_names) {
std::vector<double> time_info = {0, 0, 0};
DBDetector det(FLAGS_det_model_dir, FLAGS_use_gpu, FLAGS_gpu_id,
FLAGS_gpu_mem, FLAGS_cpu_math_library_num_threads,
FLAGS_use_mkldnn, FLAGS_max_side_len, FLAGS_det_db_thresh,
FLAGS_gpu_mem, FLAGS_cpu_threads,
FLAGS_enable_mkldnn, FLAGS_max_side_len, FLAGS_det_db_thresh,
FLAGS_det_db_box_thresh, FLAGS_det_db_unclip_ratio,
FLAGS_use_polygon_score, FLAGS_visualize,
FLAGS_use_tensorrt, FLAGS_use_fp16);
auto start = std::chrono::system_clock::now();
FLAGS_use_tensorrt, FLAGS_precision);
for (int i = 0; i < cv_all_img_names.size(); ++i) {
LOG(INFO) << "The predict img: " << cv_all_img_names[i];
......@@ -167,46 +99,38 @@ int main_det(int argc, char **argv) {
exit(1);
}
std::vector<std::vector<std::vector<int>>> boxes;
std::vector<double> det_times;
det.Run(srcimg, boxes);
auto end = std::chrono::system_clock::now();
auto duration =
std::chrono::duration_cast<std::chrono::microseconds>(end - start);
std::cout << "Cost "
<< double(duration.count()) *
std::chrono::microseconds::period::num /
std::chrono::microseconds::period::den
<< "s" << std::endl;
det.Run(srcimg, boxes, &det_times);
time_info[0] += det_times[0];
time_info[1] += det_times[1];
time_info[2] += det_times[2];
}
if (FLAGS_benchmark) {
AutoLogger autolog("ocr_det",
FLAGS_use_gpu,
FLAGS_use_tensorrt,
FLAGS_enable_mkldnn,
FLAGS_cpu_threads,
1,
"dynamic",
FLAGS_precision,
time_info,
cv_all_img_names.size());
autolog.report();
}
return 0;
}
int main_rec(int argc, char **argv) {
// Parsing command-line
google::ParseCommandLineFlags(&argc, &argv, true);
if (FLAGS_rec_model_dir.empty() || FLAGS_image_dir.empty()) {
std::cout << "Usage[rec]: ./ppocr --rec_model_dir=/PATH/TO/REC_INFERENCE_MODEL/ "
<< "--image_dir=/PATH/TO/INPUT/IMAGE/" << std::endl;
exit(1);
}
if (!PathExists(FLAGS_image_dir)) {
std::cerr << "[ERROR] image path not exist! image_dir: " << FLAGS_image_dir << endl;
exit(1);
}
std::vector<cv::String> cv_all_img_names;
cv::glob(FLAGS_image_dir, cv_all_img_names);
std::cout << "total images num: " << cv_all_img_names.size() << endl;
int main_rec(std::vector<cv::String> cv_all_img_names) {
std::vector<double> time_info = {0, 0, 0};
CRNNRecognizer rec(FLAGS_rec_model_dir, FLAGS_use_gpu, FLAGS_gpu_id,
FLAGS_gpu_mem, FLAGS_cpu_math_library_num_threads,
FLAGS_use_mkldnn, FLAGS_char_list_file,
FLAGS_use_tensorrt, FLAGS_use_fp16);
auto start = std::chrono::system_clock::now();
FLAGS_gpu_mem, FLAGS_cpu_threads,
FLAGS_enable_mkldnn, FLAGS_char_list_file,
FLAGS_use_tensorrt, FLAGS_precision);
for (int i = 0; i < cv_all_img_names.size(); ++i) {
LOG(INFO) << "The predict img: " << cv_all_img_names[i];
......@@ -217,65 +141,38 @@ int main_rec(int argc, char **argv) {
exit(1);
}
rec.Run(srcimg);
std::vector<double> rec_times;
rec.Run(srcimg, &rec_times);
auto end = std::chrono::system_clock::now();
auto duration =
std::chrono::duration_cast<std::chrono::microseconds>(end - start);
std::cout << "Cost "
<< double(duration.count()) *
std::chrono::microseconds::period::num /
std::chrono::microseconds::period::den
<< "s" << std::endl;
time_info[0] += rec_times[0];
time_info[1] += rec_times[1];
time_info[2] += rec_times[2];
}
return 0;
}
int main_system(int argc, char **argv) {
// Parsing command-line
google::ParseCommandLineFlags(&argc, &argv, true);
if ((FLAGS_det_model_dir.empty() || FLAGS_rec_model_dir.empty() || FLAGS_image_dir.empty()) ||
(FLAGS_use_angle_cls && FLAGS_cls_model_dir.empty())) {
std::cout << "Usage[system without angle cls]: ./ppocr --det_model_dir=/PATH/TO/DET_INFERENCE_MODEL/ "
<< "--rec_model_dir=/PATH/TO/REC_INFERENCE_MODEL/ "
<< "--image_dir=/PATH/TO/INPUT/IMAGE/" << std::endl;
std::cout << "Usage[system with angle cls]: ./ppocr --det_model_dir=/PATH/TO/DET_INFERENCE_MODEL/ "
<< "--use_angle_cls=true "
<< "--cls_model_dir=/PATH/TO/CLS_INFERENCE_MODEL/ "
<< "--rec_model_dir=/PATH/TO/REC_INFERENCE_MODEL/ "
<< "--image_dir=/PATH/TO/INPUT/IMAGE/" << std::endl;
exit(1);
}
if (!PathExists(FLAGS_image_dir)) {
std::cerr << "[ERROR] image path not exist! image_dir: " << FLAGS_image_dir << endl;
exit(1);
}
std::vector<cv::String> cv_all_img_names;
cv::glob(FLAGS_image_dir, cv_all_img_names);
std::cout << "total images num: " << cv_all_img_names.size() << endl;
int main_system(std::vector<cv::String> cv_all_img_names) {
DBDetector det(FLAGS_det_model_dir, FLAGS_use_gpu, FLAGS_gpu_id,
FLAGS_gpu_mem, FLAGS_cpu_math_library_num_threads,
FLAGS_use_mkldnn, FLAGS_max_side_len, FLAGS_det_db_thresh,
FLAGS_gpu_mem, FLAGS_cpu_threads,
FLAGS_enable_mkldnn, FLAGS_max_side_len, FLAGS_det_db_thresh,
FLAGS_det_db_box_thresh, FLAGS_det_db_unclip_ratio,
FLAGS_use_polygon_score, FLAGS_visualize,
FLAGS_use_tensorrt, FLAGS_use_fp16);
FLAGS_use_tensorrt, FLAGS_precision);
Classifier *cls = nullptr;
if (FLAGS_use_angle_cls) {
cls = new Classifier(FLAGS_cls_model_dir, FLAGS_use_gpu, FLAGS_gpu_id,
FLAGS_gpu_mem, FLAGS_cpu_math_library_num_threads,
FLAGS_use_mkldnn, FLAGS_cls_thresh,
FLAGS_use_tensorrt, FLAGS_use_fp16);
FLAGS_gpu_mem, FLAGS_cpu_threads,
FLAGS_enable_mkldnn, FLAGS_cls_thresh,
FLAGS_use_tensorrt, FLAGS_precision);
}
CRNNRecognizer rec(FLAGS_rec_model_dir, FLAGS_use_gpu, FLAGS_gpu_id,
FLAGS_gpu_mem, FLAGS_cpu_math_library_num_threads,
FLAGS_use_mkldnn, FLAGS_char_list_file,
FLAGS_use_tensorrt, FLAGS_use_fp16);
FLAGS_gpu_mem, FLAGS_cpu_threads,
FLAGS_enable_mkldnn, FLAGS_char_list_file,
FLAGS_use_tensorrt, FLAGS_precision);
auto start = std::chrono::system_clock::now();
......@@ -288,17 +185,19 @@ int main_system(int argc, char **argv) {
exit(1);
}
std::vector<std::vector<std::vector<int>>> boxes;
det.Run(srcimg, boxes);
std::vector<double> det_times;
std::vector<double> rec_times;
det.Run(srcimg, boxes, &det_times);
cv::Mat crop_img;
for (int j = 0; j < boxes.size(); j++) {
crop_img = GetRotateCropImage(srcimg, boxes[j]);
crop_img = Utility::GetRotateCropImage(srcimg, boxes[j]);
if (cls != nullptr) {
crop_img = cls->Run(crop_img);
}
rec.Run(crop_img);
rec.Run(crop_img, &rec_times);
}
auto end = std::chrono::system_clock::now();
......@@ -315,22 +214,70 @@ int main_system(int argc, char **argv) {
}
void check_params(char* mode) {
if (strcmp(mode, "det")==0) {
if (FLAGS_det_model_dir.empty() || FLAGS_image_dir.empty()) {
std::cout << "Usage[det]: ./ppocr --det_model_dir=/PATH/TO/DET_INFERENCE_MODEL/ "
<< "--image_dir=/PATH/TO/INPUT/IMAGE/" << std::endl;
exit(1);
}
}
if (strcmp(mode, "rec")==0) {
if (FLAGS_rec_model_dir.empty() || FLAGS_image_dir.empty()) {
std::cout << "Usage[rec]: ./ppocr --rec_model_dir=/PATH/TO/REC_INFERENCE_MODEL/ "
<< "--image_dir=/PATH/TO/INPUT/IMAGE/" << std::endl;
exit(1);
}
}
if (strcmp(mode, "system")==0) {
if ((FLAGS_det_model_dir.empty() || FLAGS_rec_model_dir.empty() || FLAGS_image_dir.empty()) ||
(FLAGS_use_angle_cls && FLAGS_cls_model_dir.empty())) {
std::cout << "Usage[system without angle cls]: ./ppocr --det_model_dir=/PATH/TO/DET_INFERENCE_MODEL/ "
<< "--rec_model_dir=/PATH/TO/REC_INFERENCE_MODEL/ "
<< "--image_dir=/PATH/TO/INPUT/IMAGE/" << std::endl;
std::cout << "Usage[system with angle cls]: ./ppocr --det_model_dir=/PATH/TO/DET_INFERENCE_MODEL/ "
<< "--use_angle_cls=true "
<< "--cls_model_dir=/PATH/TO/CLS_INFERENCE_MODEL/ "
<< "--rec_model_dir=/PATH/TO/REC_INFERENCE_MODEL/ "
<< "--image_dir=/PATH/TO/INPUT/IMAGE/" << std::endl;
exit(1);
}
}
if (FLAGS_precision != "fp32" && FLAGS_precision != "fp16" && FLAGS_precision != "int8") {
cout << "precison should be 'fp32'(default), 'fp16' or 'int8'. " << endl;
exit(1);
}
}
int main(int argc, char **argv) {
if (strcmp(argv[1], "det")!=0 && strcmp(argv[1], "rec")!=0 && strcmp(argv[1], "system")!=0) {
std::cout << "Please choose one mode of [det, rec, system] !" << std::endl;
return -1;
}
std::cout << "mode: " << argv[1] << endl;
if (strcmp(argv[1], "det")==0) {
return main_det(argc, argv);
}
if (strcmp(argv[1], "rec")==0) {
return main_rec(argc, argv);
}
if (strcmp(argv[1], "system")==0) {
return main_system(argc, argv);
}
if (argc<=1 || (strcmp(argv[1], "det")!=0 && strcmp(argv[1], "rec")!=0 && strcmp(argv[1], "system")!=0)) {
std::cout << "Please choose one mode of [det, rec, system] !" << std::endl;
return -1;
}
std::cout << "mode: " << argv[1] << endl;
// Parsing command-line
google::ParseCommandLineFlags(&argc, &argv, true);
check_params(argv[1]);
if (!PathExists(FLAGS_image_dir)) {
std::cerr << "[ERROR] image path not exist! image_dir: " << FLAGS_image_dir << endl;
exit(1);
}
// return 0;
std::vector<cv::String> cv_all_img_names;
cv::glob(FLAGS_image_dir, cv_all_img_names);
std::cout << "total images num: " << cv_all_img_names.size() << endl;
if (strcmp(argv[1], "det")==0) {
return main_det(cv_all_img_names);
}
if (strcmp(argv[1], "rec")==0) {
return main_rec(cv_all_img_names);
}
if (strcmp(argv[1], "system")==0) {
return main_system(cv_all_img_names);
}
}
deploy/cpp_infer/src/ocr_cls.cpp
浏览文件 @ a8a9b2e5
......@@ -77,10 +77,16 @@ void Classifier::LoadModel(const std::string &model_dir) {
if (this->use_gpu_) {
config.EnableUseGpu(this->gpu_mem_, this->gpu_id_);
if (this->use_tensorrt_) {
auto precision = paddle_infer::Config::Precision::kFloat32;
if (this->precision_ == "fp16") {
precision = paddle_infer::Config::Precision::kHalf;
}
if (this->precision_ == "int8") {
precision = paddle_infer::Config::Precision::kInt8;
}
config.EnableTensorRtEngine(
1 << 20, 10, 3,
this->use_fp16_ ? paddle_infer::Config::Precision::kHalf
: paddle_infer::Config::Precision::kFloat32,
precision,
false, false);
}
} else {
......
deploy/cpp_infer/src/ocr_det.cpp
浏览文件 @ a8a9b2e5
......@@ -26,10 +26,16 @@ void DBDetector::LoadModel(const std::string &model_dir) {
if (this->use_gpu_) {
config.EnableUseGpu(this->gpu_mem_, this->gpu_id_);
if (this->use_tensorrt_) {
auto precision = paddle_infer::Config::Precision::kFloat32;
if (this->precision_ == "fp16") {
precision = paddle_infer::Config::Precision::kHalf;
}
if (this->precision_ == "int8") {
precision = paddle_infer::Config::Precision::kInt8;
}
config.EnableTensorRtEngine(
1 << 20, 10, 3,
this->use_fp16_ ? paddle_infer::Config::Precision::kHalf
: paddle_infer::Config::Precision::kFloat32,
precision,
false, false);
std::map<std::string, std::vector<int>> min_input_shape = {
{"x", {1, 3, 50, 50}},
......@@ -91,13 +97,16 @@ void DBDetector::LoadModel(const std::string &model_dir) {
}
void DBDetector::Run(cv::Mat &img,
std::vector<std::vector<std::vector<int>>> &boxes) {
std::vector<std::vector<std::vector<int>>> &boxes,
std::vector<double> *times) {
float ratio_h{};
float ratio_w{};
cv::Mat srcimg;
cv::Mat resize_img;
img.copyTo(srcimg);
auto preprocess_start = std::chrono::steady_clock::now();
this->resize_op_.Run(img, resize_img, this->max_side_len_, ratio_h, ratio_w,
this->use_tensorrt_);
......@@ -106,14 +115,17 @@ void DBDetector::Run(cv::Mat &img,
std::vector<float> input(1 * 3 * resize_img.rows * resize_img.cols, 0.0f);
this->permute_op_.Run(&resize_img, input.data());
auto preprocess_end = std::chrono::steady_clock::now();
// Inference.
auto input_names = this->predictor_->GetInputNames();
auto input_t = this->predictor_->GetInputHandle(input_names[0]);
input_t->Reshape({1, 3, resize_img.rows, resize_img.cols});
auto inference_start = std::chrono::steady_clock::now();
input_t->CopyFromCpu(input.data());
this->predictor_->Run();
std::vector<float> out_data;
auto output_names = this->predictor_->GetOutputNames();
auto output_t = this->predictor_->GetOutputHandle(output_names[0]);
......@@ -123,7 +135,9 @@ void DBDetector::Run(cv::Mat &img,
out_data.resize(out_num);
output_t->CopyToCpu(out_data.data());
auto inference_end = std::chrono::steady_clock::now();
auto postprocess_start = std::chrono::steady_clock::now();
int n2 = output_shape[2];
int n3 = output_shape[3];
int n = n2 * n3;
......@@ -151,7 +165,15 @@ void DBDetector::Run(cv::Mat &img,
this->det_db_unclip_ratio_, this->use_polygon_score_);
boxes = post_processor_.FilterTagDetRes(boxes, ratio_h, ratio_w, srcimg);
auto postprocess_end = std::chrono::steady_clock::now();
std::cout << "Detected boxes num: " << boxes.size() << endl;
std::chrono::duration<float> preprocess_diff = preprocess_end - preprocess_start;
times->push_back(double(preprocess_diff.count() * 1000));
std::chrono::duration<float> inference_diff = inference_end - inference_start;
times->push_back(double(inference_diff.count() * 1000));
std::chrono::duration<float> postprocess_diff = postprocess_end - postprocess_start;
times->push_back(double(postprocess_diff.count() * 1000));
//// visualization
if (this->visualize_) {
......
deploy/cpp_infer/src/ocr_rec.cpp
浏览文件 @ a8a9b2e5
......@@ -16,13 +16,13 @@
namespace PaddleOCR {
void CRNNRecognizer::Run(cv::Mat &img) {
void CRNNRecognizer::Run(cv::Mat &img, std::vector<double> *times) {
cv::Mat srcimg;
img.copyTo(srcimg);
cv::Mat resize_img;
float wh_ratio = float(srcimg.cols) / float(srcimg.rows);
auto preprocess_start = std::chrono::steady_clock::now();
this->resize_op_.Run(srcimg, resize_img, wh_ratio, this->use_tensorrt_);
this->normalize_op_.Run(&resize_img, this->mean_, this->scale_,
......@@ -31,11 +31,13 @@ void CRNNRecognizer::Run(cv::Mat &img) {
std::vector<float> input(1 * 3 * resize_img.rows * resize_img.cols, 0.0f);
this->permute_op_.Run(&resize_img, input.data());
auto preprocess_end = std::chrono::steady_clock::now();
// Inference.
auto input_names = this->predictor_->GetInputNames();
auto input_t = this->predictor_->GetInputHandle(input_names[0]);
input_t->Reshape({1, 3, resize_img.rows, resize_img.cols});
auto inference_start = std::chrono::steady_clock::now();
input_t->CopyFromCpu(input.data());
this->predictor_->Run();
......@@ -49,8 +51,10 @@ void CRNNRecognizer::Run(cv::Mat &img) {
predict_batch.resize(out_num);
output_t->CopyToCpu(predict_batch.data());
auto inference_end = std::chrono::steady_clock::now();
// ctc decode
auto postprocess_start = std::chrono::steady_clock::now();
std::vector<std::string> str_res;
int argmax_idx;
int last_index = 0;
......@@ -73,11 +77,19 @@ void CRNNRecognizer::Run(cv::Mat &img) {
}
last_index = argmax_idx;
}
auto postprocess_end = std::chrono::steady_clock::now();
score /= count;
for (int i = 0; i < str_res.size(); i++) {
std::cout << str_res[i];
}
std::cout << "\tscore: " << score << std::endl;
std::chrono::duration<float> preprocess_diff = preprocess_end - preprocess_start;
times->push_back(double(preprocess_diff.count() * 1000));
std::chrono::duration<float> inference_diff = inference_end - inference_start;
times->push_back(double(inference_diff.count() * 1000));
std::chrono::duration<float> postprocess_diff = postprocess_end - postprocess_start;
times->push_back(double(postprocess_diff.count() * 1000));
}
void CRNNRecognizer::LoadModel(const std::string &model_dir) {
......@@ -89,10 +101,16 @@ void CRNNRecognizer::LoadModel(const std::string &model_dir) {
if (this->use_gpu_) {
config.EnableUseGpu(this->gpu_mem_, this->gpu_id_);
if (this->use_tensorrt_) {
auto precision = paddle_infer::Config::Precision::kFloat32;
if (this->precision_ == "fp16") {
precision = paddle_infer::Config::Precision::kHalf;
}
if (this->precision_ == "int8") {
precision = paddle_infer::Config::Precision::kInt8;
}
config.EnableTensorRtEngine(
1 << 20, 10, 3,
this->use_fp16_ ? paddle_infer::Config::Precision::kHalf
: paddle_infer::Config::Precision::kFloat32,
precision,
false, false);
std::map<std::string, std::vector<int>> min_input_shape = {
{"x", {1, 3, 32, 10}}};
......@@ -126,59 +144,4 @@ void CRNNRecognizer::LoadModel(const std::string &model_dir) {
this->predictor_ = CreatePredictor(config);
}
cv::Mat CRNNRecognizer::GetRotateCropImage(const cv::Mat &srcimage,
std::vector<std::vector<int>> box) {
cv::Mat image;
srcimage.copyTo(image);
std::vector<std::vector<int>> points = box;
int x_collect[4] = {box[0][0], box[1][0], box[2][0], box[3][0]};
int y_collect[4] = {box[0][1], box[1][1], box[2][1], box[3][1]};
int left = int(*std::min_element(x_collect, x_collect + 4));
int right = int(*std::max_element(x_collect, x_collect + 4));
int top = int(*std::min_element(y_collect, y_collect + 4));
int bottom = int(*std::max_element(y_collect, y_collect + 4));
cv::Mat img_crop;
image(cv::Rect(left, top, right - left, bottom - top)).copyTo(img_crop);
for (int i = 0; i < points.size(); i++) {
points[i][0] -= left;
points[i][1] -= top;
}
int img_crop_width = int(sqrt(pow(points[0][0] - points[1][0], 2) +
pow(points[0][1] - points[1][1], 2)));
int img_crop_height = int(sqrt(pow(points[0][0] - points[3][0], 2) +
pow(points[0][1] - points[3][1], 2)));
cv::Point2f pts_std[4];
pts_std[0] = cv::Point2f(0., 0.);
pts_std[1] = cv::Point2f(img_crop_width, 0.);
pts_std[2] = cv::Point2f(img_crop_width, img_crop_height);
pts_std[3] = cv::Point2f(0.f, img_crop_height);
cv::Point2f pointsf[4];
pointsf[0] = cv::Point2f(points[0][0], points[0][1]);
pointsf[1] = cv::Point2f(points[1][0], points[1][1]);
pointsf[2] = cv::Point2f(points[2][0], points[2][1]);
pointsf[3] = cv::Point2f(points[3][0], points[3][1]);
cv::Mat M = cv::getPerspectiveTransform(pointsf, pts_std);
cv::Mat dst_img;
cv::warpPerspective(img_crop, dst_img, M,
cv::Size(img_crop_width, img_crop_height),
cv::BORDER_REPLICATE);
if (float(dst_img.rows) >= float(dst_img.cols) * 1.5) {
cv::Mat srcCopy = cv::Mat(dst_img.rows, dst_img.cols, dst_img.depth());
cv::transpose(dst_img, srcCopy);
cv::flip(srcCopy, srcCopy, 0);
return srcCopy;
} else {
return dst_img;
}
}
} // namespace PaddleOCR
deploy/cpp_infer/src/utility.cpp
浏览文件 @ a8a9b2e5
......@@ -92,4 +92,59 @@ void Utility::GetAllFiles(const char *dir_name,
}
}
cv::Mat Utility::GetRotateCropImage(const cv::Mat &srcimage,
std::vector<std::vector<int>> box) {
cv::Mat image;
srcimage.copyTo(image);
std::vector<std::vector<int>> points = box;
int x_collect[4] = {box[0][0], box[1][0], box[2][0], box[3][0]};
int y_collect[4] = {box[0][1], box[1][1], box[2][1], box[3][1]};
int left = int(*std::min_element(x_collect, x_collect + 4));
int right = int(*std::max_element(x_collect, x_collect + 4));
int top = int(*std::min_element(y_collect, y_collect + 4));
int bottom = int(*std::max_element(y_collect, y_collect + 4));
cv::Mat img_crop;
image(cv::Rect(left, top, right - left, bottom - top)).copyTo(img_crop);
for (int i = 0; i < points.size(); i++) {
points[i][0] -= left;
points[i][1] -= top;
}
int img_crop_width = int(sqrt(pow(points[0][0] - points[1][0], 2) +
pow(points[0][1] - points[1][1], 2)));
int img_crop_height = int(sqrt(pow(points[0][0] - points[3][0], 2) +
pow(points[0][1] - points[3][1], 2)));
cv::Point2f pts_std[4];
pts_std[0] = cv::Point2f(0., 0.);
pts_std[1] = cv::Point2f(img_crop_width, 0.);
pts_std[2] = cv::Point2f(img_crop_width, img_crop_height);
pts_std[3] = cv::Point2f(0.f, img_crop_height);
cv::Point2f pointsf[4];
pointsf[0] = cv::Point2f(points[0][0], points[0][1]);
pointsf[1] = cv::Point2f(points[1][0], points[1][1]);
pointsf[2] = cv::Point2f(points[2][0], points[2][1]);
pointsf[3] = cv::Point2f(points[3][0], points[3][1]);
cv::Mat M = cv::getPerspectiveTransform(pointsf, pts_std);
cv::Mat dst_img;
cv::warpPerspective(img_crop, dst_img, M,
cv::Size(img_crop_width, img_crop_height),
cv::BORDER_REPLICATE);
if (float(dst_img.rows) >= float(dst_img.cols) * 1.5) {
cv::Mat srcCopy = cv::Mat(dst_img.rows, dst_img.cols, dst_img.depth());
cv::transpose(dst_img, srcCopy);
cv::flip(srcCopy, srcCopy, 0);
return srcCopy;
} else {
return dst_img;
}
}
} // namespace PaddleOCR
\ No newline at end of file
deploy/slim/prune/sensitivity_anal.py
浏览文件 @ a8a9b2e5
......@@ -75,7 +75,7 @@ def main(config, device, logger, vdl_writer):
model = build_model(config['Architecture'])
flops = paddle.flops(model, [1, 3, 640, 640])
logger.info(f"FLOPs before pruning: {flops}")
logger.info("FLOPs before pruning: {}".format(flops))
from paddleslim.dygraph import FPGMFilterPruner
model.train()
......@@ -106,8 +106,8 @@ def main(config, device, logger, vdl_writer):
def eval_fn():
metric = program.eval(model, valid_dataloader, post_process_class,
eval_class)
logger.info(f"metric['hmean']: {metric['hmean']}")
eval_class, False)
logger.info("metric['hmean']: {}".format(metric['hmean']))
return metric['hmean']
params_sensitive = pruner.sensitive(
......@@ -123,16 +123,17 @@ def main(config, device, logger, vdl_writer):
# calculate pruned params's ratio
params_sensitive = pruner._get_ratios_by_loss(params_sensitive, loss=0.02)
for key in params_sensitive.keys():
logger.info(f"{key}, {params_sensitive[key]}")
logger.info("{}, {}".format(key, params_sensitive[key]))
#params_sensitive = {}
#for param in model.parameters():
# if 'transpose' not in param.name and 'linear' not in param.name:
# params_sensitive[param.name] = 0.1
plan = pruner.prune_vars(params_sensitive, [0])
for param in model.parameters():
if ("weights" in param.name and "conv" in param.name) or (
"w_0" in param.name and "conv2d" in param.name):
logger.info(f"{param.name}: {param.shape}")
flops = paddle.flops(model, [1, 3, 640, 640])
logger.info(f"FLOPs after pruning: {flops}")
logger.info("FLOPs after pruning: {}".format(flops))
# start train
......
doc/doc_ch/FAQ.md
浏览文件 @ a8a9b2e5
此差异已折叠。
doc/doc_ch/algorithm_overview.md
浏览文件 @ a8a9b2e5
......@@ -44,6 +44,7 @@ PaddleOCR基于动态图开源的文本识别算法列表：
- [x] STAR-Net([paper](http://www.bmva.org/bmvc/2016/papers/paper043/index.html))[11]
- [x] RARE([paper](https://arxiv.org/abs/1603.03915v1))[12]
- [x] SRN([paper](https://arxiv.org/abs/2003.12294))[5]
- [x] NRTR([paper](https://arxiv.org/abs/1806.00926v2))
参考[DTRB][3](https://arxiv.org/abs/1904.01906)文字识别训练和评估流程，使用MJSynth和SynthText两个文字识别数据集训练，在IIIT, SVT, IC03, IC13, IC15, SVTP, CUTE数据集上进行评估，算法效果如下：
......@@ -58,6 +59,7 @@ PaddleOCR基于动态图开源的文本识别算法列表：
|RARE|MobileNetV3|82.5%|rec_mv3_tps_bilstm_att |[下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_mv3_tps_bilstm_att_v2.0_train.tar)|
|RARE|Resnet34_vd|83.6%|rec_r34_vd_tps_bilstm_att |[下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_r34_vd_tps_bilstm_att_v2.0_train.tar)|
|SRN|Resnet50_vd_fpn| 88.52% | rec_r50fpn_vd_none_srn | [下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_r50_vd_srn_train.tar) |
|NRTR|NRTR_MTB| 84.3% | rec_mtb_nrtr | [下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_mtb_nrtr_train.tar) |
PaddleOCR文本识别算法的训练和使用请参考文档教程中[模型训练/评估中的文本识别部分](./recognition.md)。
doc/doc_ch/config.md
浏览文件 @ a8a9b2e5
## 可选参数列表
# 配置文件内容与生成
## 1. 可选参数列表
以下列表可以通过`--help`查看
......@@ -8,10 +10,10 @@
| -o | ALL | 设置配置文件里的参数内容 | None | 使用-o配置相较于-c选择的配置文件具有更高的优先级。例如：`-o Global.use_gpu=false` |
## 配置文件参数介绍
## 2. 配置文件参数介绍
以 `rec_chinese_lite_train_v2.0.yml ` 为例
### Global
### 2.1 Global
| 字段 | 用途 | 默认值 | 备注 |
| :----------------------: | :---------------------: | :--------------: | :--------------------: |
......@@ -52,7 +54,7 @@
### Architecture ([ppocr/modeling](../../ppocr/modeling))
在ppocr中，网络被划分为Transform,Backbone,Neck和Head四个阶段
在PaddleOCR中，网络被划分为Transform,Backbone,Neck和Head四个阶段
| 字段 | 用途 | 默认值 | 备注 |
| :---------------------: | :---------------------: | :--------------: | :--------------------: |
......@@ -121,3 +123,97 @@
| batch_size_per_card | 训练时单卡batch size | 256 | \ |
| drop_last | 是否丢弃因数据集样本数不能被 batch_size 整除而产生的最后一个不完整的mini-batch | True | \ |
| num_workers | 用于加载数据的子进程个数，若为0即为不开启子进程，在主进程中进行数据加载 | 8 | \ |
## 3. 多语言配置文件生成
PaddleOCR目前已支持80种（除中文外）语种识别，`configs/rec/multi_languages` 路径下提供了一个多语言的配置文件模版: [rec_multi_language_lite_train.yml](../../configs/rec/multi_language/rec_multi_language_lite_train.yml)。
您有两种方式创建所需的配置文件：
1. 通过脚本自动生成
[generate_multi_language_configs.py](../../configs/rec/multi_language/generate_multi_language_configs.py) 可以帮助您生成多语言模型的配置文件
- 以意大利语为例，如果您的数据是按如下格式准备的：
```
|-train_data
|- it_train.txt # 训练集标签
|- it_val.txt # 验证集标签
|- data
|- word_001.jpg
|- word_002.jpg
|- word_003.jpg
| ...
```
可以使用默认参数，生成配置文件：
```bash
# 该代码需要在指定目录运行
cd PaddleOCR/configs/rec/multi_language/
# 通过-l或者--language参数设置需要生成的语种的配置文件，该命令会将默认参数写入配置文件
python3 generate_multi_language_configs.py -l it
```
- 如果您的数据放置在其他位置，或希望使用自己的字典，可以通过指定相关参数来生成配置文件:
```bash
# -l或者--language字段是必须的
# --train修改训练集，--val修改验证集，--data_dir修改数据集目录，--dict修改字典路径， -o修改对应默认参数
cd PaddleOCR/configs/rec/multi_language/
python3 generate_multi_language_configs.py -l it \ # 语种
--train {path/of/train_label.txt} \ # 训练标签文件的路径
--val {path/of/val_label.txt} \ # 验证集标签文件的路径
--data_dir {train_data/path} \ # 训练数据的根目录
--dict {path/of/dict} \ # 字典文件路径
-o Global.use_gpu=False # 是否使用gpu
...
```
意大利文由拉丁字母组成，因此执行完命令后会得到名为 rec_latin_lite_train.yml 的配置文件。
2. 手动修改配置文件
您也可以手动修改模版中的以下几个字段得到配置文件:
```
Global:
use_gpu: True
epoch_num: 500
...
character_type: it # 需要识别的语种
character_dict_path: {path/of/dict} # 字典文件所在路径
Train:
dataset:
name: SimpleDataSet
data_dir: train_data/ # 数据存放根目录
label_file_list: ["./train_data/train_list.txt"] # 训练集label路径
...
Eval:
dataset:
name: SimpleDataSet
data_dir: train_data/ # 数据存放根目录
label_file_list: ["./train_data/val_list.txt"] # 验证集label路径
...
```
目前PaddleOCR支持的多语言算法有：
| 配置文件 | 算法名称 | backbone | trans | seq | pred | language | character_type |
| :--------: | :-------: | :-------: | :-------: | :-----: | :-----: | :-----: | :-----: |
| rec_chinese_cht_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 中文繁体 | chinese_cht|
| rec_en_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 英语（区分大小写） | EN |
| rec_french_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 法语 | french |
| rec_ger_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 德语 | german |
| rec_japan_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 日语 | japan |
| rec_korean_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 韩语 | korean |
| rec_latin_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 拉丁字母 | latin |
| rec_arabic_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 阿拉伯字母 | ar |
| rec_cyrillic_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 斯拉夫字母 | cyrillic |
| rec_devanagari_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 梵文字母 | devanagari |
更多支持语种请参考: [多语言模型](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.1/doc/doc_ch/multi_languages.md#%E8%AF%AD%E7%A7%8D%E7%BC%A9%E5%86%99)
doc/doc_ch/detection.md
浏览文件 @ a8a9b2e5
# 文字检测
本节以icdar2015数据集为例，介绍PaddleOCR中检测模型的训练、评估与测试。
# 目录
- [1. 文字检测](#1-----)
* [1.1 数据准备](#11-----)
* [1.2 下载预训练模型](#12--------)
* [1.3 启动训练](#13-----)
* [1.4 断点训练](#14-----)
* [1.5 更换Backbone 训练](#15---backbone---)
* [1.6 指标评估](#16-----)
* [1.7 测试检测效果](#17-------)
* [1.8 转inference模型测试](#18--inference----)
- [2. FAQ](#2-faq)
## 数据准备
<a name="1-----"></a>
# 1. 文字检测
本节以icdar2015数据集为例，介绍PaddleOCR中检测模型训练、评估、测试的使用方式。
<a name="11-----"></a>
## 1.1 数据准备
icdar2015 TextLocalization数据集是文本检测的数据集，包含1000张训练图像和500张测试图像。
icdar2015数据集可以从[官网](https://rrc.cvc.uab.es/?ch=4&com=downloads)下载到，首次下载需注册。
注册完成登陆后，下载下图中红色框标出的部分，其中， `Training Set Images`下载的内容保存为`icdar_c4_train_imgs`文件夹下，`Test Set Images` 下载的内容保存为`ch4_test_images`文件夹下
<p align="center">
<img src="../datasets/ic15_location_download.png" align="middle" width = "700"/>
<p align="center">
将下载到的数据集解压到工作目录下，假设解压在 PaddleOCR/train_data/ 下。另外，PaddleOCR将零散的标注文件整理成单独的标注文件
将下载到的数据集解压到工作目录下，假设解压在 PaddleOCR/train_data/下。另外，PaddleOCR将零散的标注文件整理成单独的标注文件
，您可以通过wget的方式进行下载。
```shell
# 在PaddleOCR路径下
......@@ -23,7 +46,7 @@ python gen_label.py --mode="det" --root_path="/path/to/icdar_c4_train_imgs/" \
--output_label="/path/to/train_icdar2015_label.txt"
```
解压数据集和下载标注文件后，PaddleOCR/train_data/ 有两个文件夹和两个文件，分别是：
解压数据集和下载标注文件后，PaddleOCR/train_data/ 有两个文件夹和两个文件，按照如下方式组织icdar2015数据集：
```
/PaddleOCR/train_data/icdar2015/text_localization/
└─ icdar_c4_train_imgs/ icdar数据集的训练数据
......@@ -42,11 +65,13 @@ json.dumps编码前的图像标注信息是包含多个字典的list，字典中
如果您想在其他数据集上训练，可以按照上述形式构建标注文件。
## 快速启动训练
<a name="12--------"></a>
## 1.2 下载预训练模型
首先下载模型backbone的pretrain model，PaddleOCR的检测模型目前支持两种backbone，分别是MobileNetV3、ResNet_vd系列，
您可以根据需求使用[PaddleClas](https://github.com/PaddlePaddle/PaddleClas/tree/develop/ppcls/modeling/architectures)中的模型更换backbone，
对应的backbone预训练模型可以从[PaddleClas repo 主页中找到下载链接](https://github.com/PaddlePaddle/PaddleClas#mobile-series)。
您可以根据需求使用[PaddleClas](https://github.com/PaddlePaddle/PaddleClas/tree/release/2.0/ppcls/modeling/architectures)中的模型更换backbone，
对应的backbone预训练模型可以从[PaddleClas repo 主页中找到下载链接](https://github.com/PaddlePaddle/PaddleClas/blob/release%2F2.0/README_cn.md#resnet%E5%8F%8A%E5%85%B6vd%E7%B3%BB%E5%88%97)。
```shell
cd PaddleOCR/
# 根据backbone的不同选择下载对应的预训练模型
......@@ -56,23 +81,23 @@ wget -P ./pretrain_models/ https://paddle-imagenet-models-name.bj.bcebos.com/dyg
wget -P ./pretrain_models/ https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet18_vd_pretrained.pdparams
# 或，下载ResNet50_vd的预训练模型
wget -P ./pretrain_models/ https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet50_vd_ssld_pretrained.pdparams
```
#### 启动训练
<a name="13-----"></a>
## 1.3 启动训练
*如果您安装的是cpu版本，请将配置文件中的 `use_gpu` 字段修改为false*
```shell
# 单机单卡训练 mv3_db 模型
python3 tools/train.py -c configs/det/det_mv3_db.yml \
-o Global.pretrain_weights=./pretrain_models/MobileNetV3_large_x0_5_pretrained/
-o Global.pretrain_weights=./pretrain_models/MobileNetV3_large_x0_5_pretrained
# 单机多卡训练，通过 --gpus 参数设置使用的GPU ID
python3 -m paddle.distributed.launch --gpus '0,1,2,3' tools/train.py -c configs/det/det_mv3_db.yml \
-o Global.pretrain_weights=./pretrain_models/MobileNetV3_large_x0_5_pretrained/
-o Global.pretrain_weights=./pretrain_models/MobileNetV3_large_x0_5_pretrained
```
上述指令中，通过-c 选择训练使用configs/det/det_db_mv3.yml配置文件。
有关配置文件的详细解释，请参考[链接](./config.md)。
......@@ -81,46 +106,122 @@ python3 -m paddle.distributed.launch --gpus '0,1,2,3' tools/train.py -c configs/
python3 tools/train.py -c configs/det/det_mv3_db.yml -o Optimizer.base_lr=0.0001
```
#### 断点训练
<a name="14-----"></a>
## 1.4 断点训练
如果训练程序中断，如果希望加载训练中断的模型从而恢复训练，可以通过指定Global.checkpoints指定要加载的模型路径：
```shell
python3 tools/train.py -c configs/det/det_mv3_db.yml -o Global.checkpoints=./your/trained/model
```
**注意**：`Global.checkpoints`的优先级高于`Global.pretrain_weights`的优先级，即同时指定两个参数时，优先加载`Global.checkpoints`指定的模型，如果`Global.checkpoints`指定的模型路径有误，会加载`Global.pretrain_weights`指定的模型。
<a name="15---backbone---"></a>
## 1.5 更换Backbone 训练
PaddleOCR将网络划分为四部分，分别在[ppocr/modeling](../../ppocr/modeling)下。 进入网络的数据将按照顺序(transforms->backbones->
necks->heads)依次通过这四个部分。
```bash
├── architectures # 网络的组网代码
├── transforms # 网络的图像变换模块
├── backbones # 网络的特征提取模块
├── necks # 网络的特征增强模块
└── heads # 网络的输出模块
```
如果要更换的Backbone 在PaddleOCR中有对应实现，直接修改配置yml文件中`Backbone`部分的参数即可。
**注意**：`Global.checkpoints`的优先级高于`Global.pretrain_weights`的优先级，即同时指定两个参数时，优先加载`Global.checkpoints`指定的模型，如果`Global.checkpoints`指定的模型路径有误，会加载`Global.pretrain_weights`指定的模型。
如果要使用新的Backbone，更换backbones的例子如下:
1. 在 [ppocr/modeling/backbones](../../ppocr/modeling/backbones) 文件夹下新建文件，如my_backbone.py。
2. 在 my_backbone.py 文件内添加相关代码，示例代码如下:
```python
import paddle
import paddle.nn as nn
import paddle.nn.functional as F
class MyBackbone(nn.Layer):
def __init__(self, *args, **kwargs):
super(MyBackbone, self).__init__()
# your init code
self.conv = nn.xxxx
def forward(self, inputs):
# your network forward
y = self.conv(inputs)
return y
```
3. 在 [ppocr/modeling/backbones/\__init\__.py](../../ppocr/modeling/backbones/__init__.py)文件内导入添加的`MyBackbone`模块，然后修改配置文件中Backbone进行配置即可使用，格式如下:
## 指标评估
```yaml
Backbone:
name: MyBackbone
args1: args1
```
PaddleOCR计算三个OCR检测相关的指标，分别是：Precision、Recall、Hmean。
**注意**：如果要更换网络的其他模块，可以参考[文档](./add_new_algorithm.md)。
运行如下代码，根据配置文件`det_db_mv3.yml`中`save_res_path`指定的测试集检测结果文件，计算评估指标。
<a name="16-----"></a>
## 1.6 指标评估
PaddleOCR计算三个OCR检测相关的指标，分别是：Precision、Recall、Hmean（F-Score）。
评估时设置后处理参数`box_thresh=0.5`，`unclip_ratio=1.5`，使用不同数据集、不同模型训练，可调整这两个参数进行优化
训练中模型参数默认保存在`Global.save_model_dir`目录下。在评估指标时，需要设置`Global.checkpoints`指向保存的参数文件。
```shell
python3 tools/eval.py -c configs/det/det_mv3_db.yml -o Global.checkpoints="{path/to/weights}/best_accuracy" PostProcess.box_thresh=0.5 PostProcess.unclip_ratio=1.5
python3 tools/eval.py -c configs/det/det_mv3_db.yml -o Global.checkpoints="{path/to/weights}/best_accuracy"
```
* 注：`box_thresh`、`unclip_ratio`是DB后处理所需要的参数，在评估EAST模型时不需要设置
## 测试检测效果
<a name="17-------"></a>
## 1.7 测试检测效果
测试单张图像的检测效果
```shell
python3 tools/infer_det.py -c configs/det/det_mv3_db.yml -o Global.infer_img="./doc/imgs_en/img_10.jpg" Global.pretrained_model="./output/det_db/best_accuracy"
```
测试DB模型时，调整后处理阈值，
测试DB模型时，调整后处理阈值
```shell
python3 tools/infer_det.py -c configs/det/det_mv3_db.yml -o Global.infer_img="./doc/imgs_en/img_10.jpg" Global.pretrained_model="./output/det_db/best_accuracy" PostProcess.box_thresh=0.6 PostProcess.unclip_ratio=1.5
python3 tools/infer_det.py -c configs/det/det_mv3_db.yml -o Global.infer_img="./doc/imgs_en/img_10.jpg" Global.pretrained_model="./output/det_db/best_accuracy" PostProcess.box_thresh=0.6 PostProcess.unclip_ratio=2.0
```
测试文件夹下所有图像的检测效果
```shell
python3 tools/infer_det.py -c configs/det/det_mv3_db.yml -o Global.infer_img="./doc/imgs_en/" Global.pretrained_model="./output/det_db/best_accuracy"
```
<a name="18--inference----"></a>
## 1.8 转inference模型测试
inference 模型（`paddle.jit.save`保存的模型）
一般是模型训练，把模型结构和模型参数保存在文件中的固化模型，多用于预测部署场景。
训练过程中保存的模型是checkpoints模型，保存的只有模型的参数，多用于恢复训练等。
与checkpoints模型相比，inference 模型会额外保存模型的结构信息，在预测部署、加速推理上性能优越，灵活方便，适合于实际系统集成。
检测模型转inference 模型方式：
```shell
# 加载配置文件`det_mv3_db.yml`，从`output/det_db`目录下加载`best_accuracy`模型，inference模型保存在`./output/det_db_inference`目录下
python3 tools/export_model.py -c configs/det/det_mv3_db.yml -o Global.pretrained_model="./output/det_db/best_accuracy" Global.save_inference_dir="./output/det_db_inference/"
```
DB检测模型inference 模型预测：
```shell
python3 tools/infer/predict_det.py --det_algorithm="DB" --det_model_dir="./output/det_db_inference/" --image_dir="./doc/imgs/" --use_gpu=True
```
如果是其他检测，比如EAST模型，det_algorithm参数需要修改为EAST，默认为DB算法：
```shell
python3 tools/infer/predict_det.py --det_algorithm="EAST" --det_model_dir="./output/det_db_inference/" --image_dir="./doc/imgs/" --use_gpu=True
```
<a name="2"></a>
# 2. FAQ
Q1: 训练模型转inference 模型之后预测效果不一致？
**A**：此类问题出现较多，问题多是trained model预测时候的预处理、后处理参数和inference model预测的时候的预处理、后处理参数不一致导致的。以det_mv3_db.yml配置文件训练的模型为例，训练模型、inference模型预测结果不一致问题解决方式如下：
- 检查[trained model预处理](https://github.com/PaddlePaddle/PaddleOCR/blob/c1ed243fb68d5d466258243092e56cbae32e2c14/configs/det/det_mv3_db.yml#L116)，和[inference model的预测预处理](https://github.com/PaddlePaddle/PaddleOCR/blob/c1ed243fb68d5d466258243092e56cbae32e2c14/tools/infer/predict_det.py#L42)函数是否一致。算法在评估的时候，输入图像大小会影响精度，为了和论文保持一致，训练icdar15配置文件中将图像resize到[736, 1280]，但是在inference model预测的时候只有一套默认参数，会考虑到预测速度问题，默认限制图像最长边为960做resize的。训练模型预处理和inference模型的预处理函数位于[ppocr/data/imaug/operators.py](https://github.com/PaddlePaddle/PaddleOCR/blob/c1ed243fb68d5d466258243092e56cbae32e2c14/ppocr/data/imaug/operators.py#L147)
- 检查[trained model后处理](https://github.com/PaddlePaddle/PaddleOCR/blob/c1ed243fb68d5d466258243092e56cbae32e2c14/configs/det/det_mv3_db.yml#L51)，和[inference 后处理参数](https://github.com/PaddlePaddle/PaddleOCR/blob/c1ed243fb68d5d466258243092e56cbae32e2c14/tools/infer/utility.py#L50)是否一致。
doc/doc_ch/environment.md 0 → 100644
浏览文件 @ a8a9b2e5
# 运行环境准备
Windows和Mac用户推荐使用Anaconda搭建Python环境，Linux用户建议使用docker搭建PyThon环境。
如果对于Python环境熟悉的用户可以直接跳到第2步安装PaddlePaddle。
* [1. Python环境搭建](#1)
+ [1.1 Windows](#1.1)
+ [1.2 Mac](#1.2)
+ [1.3 Linux](#1.3)
* [2. 安装PaddlePaddle](#2)
<a name="1"></a>
## 1. Python环境搭建
<a name="1.1"></a>
### 1.1 Windows
#### 1.1.1 安装Anaconda
- 说明：使用paddlepaddle需要先安装python环境，这里我们选择python集成环境Anaconda工具包
- Anaconda是1个常用的python包管理程序
- 安装完Anaconda后，可以安装python环境，以及numpy等所需的工具包环境。
- Anaconda下载：
- 地址：https://mirrors.tuna.tsinghua.edu.cn/anaconda/archive/?C=M&O=D
- 大部分win10电脑均为64位操作系统，选择x86_64版本；若电脑为32位操作系统，则选择x86.exe
<img src="../install/windows/Anaconda_download.png" alt="anaconda download" width="800" align="center"/>
- 下载完成后，双击安装程序进入图形界面
- 默认安装位置为C盘，建议将安装位置更改到D盘：
<img src="../install/windows/anaconda_install_folder.png" alt="install config" width="500" align="center"/>
- 勾选conda加入环境变量，忽略警告：
<img src="../install/windows/anaconda_install_env.png" alt="add conda to path" width="500" align="center"/>
#### 1.1.2 打开终端并创建conda环境
- 打开Anaconda Prompt终端：左下角Windows Start Menu -> Anaconda3 -> Anaconda Prompt启动控制台
<img src="../install/windows/anaconda_prompt.png" alt="anaconda download" width="300" align="center"/>
- 创建新的conda环境
```shell
# 在命令行输入以下命令，创建名为paddle_env的环境
# 此处为加速下载，使用清华源
conda create --name paddle_env python=3.8 --channel https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/ # 这是一行命令
```
该命令会创建1个名为paddle_env、python版本为3.8的可执行环境，根据网络状态，需要花费一段时间
之后命令行中会输出提示信息，输入y并回车继续安装
<img src="../install/windows/conda_new_env.png" alt="conda create" width="700" align="center"/>
- 激活刚创建的conda环境，在命令行中输入以下命令：
```shell
# 激活paddle_env环境
conda activate paddle_env
# 查看当前python的位置
where python
```
<img src="../install/windows/conda_list_env.png" alt="create environment" width="600" align="center"/>
以上anaconda环境和python环境安装完毕
<a name="1.2"></a>
### 1.2 Mac
#### 1.2.1 安装Anaconda
- 说明：使用paddlepaddle需要先安装python环境，这里我们选择python集成环境Anaconda工具包
- Anaconda是1个常用的python包管理程序
- 安装完Anaconda后，可以安装python环境，以及numpy等所需的工具包环境
- Anaconda下载：
- 地址：https://mirrors.tuna.tsinghua.edu.cn/anaconda/archive/?C=M&O=D
<img src="../install/mac/anaconda_start.png" alt="anaconda download" width="800" align="center"/>
- 选择最下方的`Anaconda3-2021.05-MacOSX-x86_64.pkg`下载
- 下载完成后，双击.pkg文件进入图形界面
- 按默认设置即可，安装需要花费一段时间
- 建议安装vscode或pycharm等代码编辑器
#### 1.2.2 打开终端并创建conda环境
- 打开终端
- 同时按下command键和空格键，在聚焦搜索中输入"终端"，双击进入终端
- **将conda加入环境变量**
- 加入环境变量是为了让系统能识别conda命令
- 输入以下命令，在终端中打开`~/.bash_profile`：
```shell
vim ~/.bash_profile
```
- 在`~/.bash_profile`中将conda添加为环境变量：
```shell
# 先按i进入编辑模式
# 在第一行输入：
export PATH="~/opt/anaconda3/bin:$PATH"
# 若安装时自定义了安装位置，则将~/opt/anaconda3/bin改为自定义的安装目录下的bin文件夹
```
```shell
# 修改后的~/.bash_profile文件应如下（其中xxx为用户名）：
export PATH="~/opt/anaconda3/bin:$PATH"
# >>> conda initialize >>>
# !! Contents within this block are managed by 'conda init' !!
__conda_setup="$('/Users/xxx/opt/anaconda3/bin/conda' 'shell.bash' 'hook' 2> /dev/null)"
if [ $? -eq 0 ]; then
eval "$__conda_setup"
else
if [ -f "/Users/xxx/opt/anaconda3/etc/profile.d/conda.sh" ]; then
. "/Users/xxx/opt/anaconda3/etc/profile.d/conda.sh"
else
export PATH="/Users/xxx/opt/anaconda3/bin:$PATH"
fi
fi
unset __conda_setup
# <<< conda initialize <<<
```
- 修改完成后，先按`esc`键退出编辑模式，再输入`:wq!`并回车，以保存退出
- 验证是否能识别conda命令：
- 在终端中输入`source ~/.bash_profile`以更新环境变量
- 再在终端输入`conda info --envs`，若能显示当前有base环境，则conda已加入环境变量
- 创建新的conda环境
```shell
# 在命令行输入以下命令，创建名为paddle_env的环境
# 此处为加速下载，使用清华源
conda create --name paddle_env python=3.8 --channel https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
```
- 该命令会创建1个名为paddle_env、python版本为3.8的可执行环境，根据网络状态，需要花费一段时间
- 之后命令行中会输出提示信息，输入y并回车继续安装
- <img src="../install/mac/conda_create.png" alt="conda_create" width="600" align="center"/>
- 激活刚创建的conda环境，在命令行中输入以下命令：
```shell
# 激活paddle_env环境
conda activate paddle_env
# 查看当前python的位置
where python
```
<img src="../install/mac/conda_activate.png" alt="conda_actviate" width="600" align="center"/>
以上anaconda环境和python环境安装完毕
<a name="1.3"></a>
### 1.3 Linux
Linux用户可选择Anaconda或Docker两种方式运行。如果你熟悉Docker且需要训练PaddleOCR模型，推荐使用Docker环境，PaddleOCR的开发流程均在Docker环境下运行。如果你不熟悉Docker，也可以使用Anaconda来运行项目。
#### 1.3.1 Anaconda环境配置
- 说明：使用paddlepaddle需要先安装python环境，这里我们选择python集成环境Anaconda工具包
- Anaconda是1个常用的python包管理程序
- 安装完Anaconda后，可以安装python环境，以及numpy等所需的工具包环境
- **下载Anaconda**：
- 下载地址：https://mirrors.tuna.tsinghua.edu.cn/anaconda/archive/?C=M&O=D
<img src="../install/linux/anaconda_download.png" akt="anaconda download" width="800" align="center"/>
- 选择适合您操作系统的版本
- 可在终端输入`uname -m`查询系统所用的指令集
- 下载法1：本地下载，再将安装包传到linux服务器上
- 下载法2：直接使用linux命令行下载
```shell
# 首先安装wget
sudo apt-get install wget # Ubuntu
sudo yum install wget # CentOS
```
```shell
# 然后使用wget从清华源上下载
# 如要下载Anaconda3-2021.05-Linux-x86_64.sh，则下载命令如下：
wget https://mirrors.tuna.tsinghua.edu.cn/anaconda/archive/Anaconda3-2021.05-Linux-x86_64.sh
# 若您要下载其他版本，需要将最后1个/后的文件名改成您希望下载的版本
```
- 安装Anaconda：
- 在命令行输入`sh Anaconda3-2021.05-Linux-x86_64.sh`
- 若您下载的是其它版本，则将该命令的文件名替换为您下载的文件名
- 按照安装提示安装即可
- 查看许可时可输入q来退出
- **将conda加入环境变量**
- 加入环境变量是为了让系统能识别conda命令，若您在安装时已将conda加入环境变量path，则可跳过本步
- 在终端中打开`~/.bashrc`：
```shell
# 在终端中输入以下命令：
vim ~/.bashrc
```
- 在`~/.bashrc`中将conda添加为环境变量：
```shell
# 先按i进入编辑模式
# 在第一行输入：
export PATH="~/anaconda3/bin:$PATH"
# 若安装时自定义了安装位置，则将~/anaconda3/bin改为自定义的安装目录下的bin文件夹
```
```shell
# 修改后的~/.bash_profile文件应如下（其中xxx为用户名）：
export PATH="~/opt/anaconda3/bin:$PATH"
# >>> conda initialize >>>
# !! Contents within this block are managed by 'conda init' !!
__conda_setup="$('/Users/xxx/opt/anaconda3/bin/conda' 'shell.bash' 'hook' 2> /dev/null)"
if [ $? -eq 0 ]; then
eval "$__conda_setup"
else
if [ -f "/Users/xxx/opt/anaconda3/etc/profile.d/conda.sh" ]; then
. "/Users/xxx/opt/anaconda3/etc/profile.d/conda.sh"
else
export PATH="/Users/xxx/opt/anaconda3/bin:$PATH"
fi
fi
unset __conda_setup
# <<< conda initialize <<<
```
- 修改完成后，先按`esc`键退出编辑模式，再输入`:wq!`并回车，以保存退出
- 验证是否能识别conda命令：
- 在终端中输入`source ~/.bash_profile`以更新环境变量
- 再在终端输入`conda info --envs`，若能显示当前有base环境，则conda已加入环境变量
- 创建新的conda环境
```shell
# 在命令行输入以下命令，创建名为paddle_env的环境
# 此处为加速下载，使用清华源
conda create --name paddle_env python=3.8 --channel https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
```
- 该命令会创建1个名为paddle_env、python版本为3.8的可执行环境，根据网络状态，需要花费一段时间
- 之后命令行中会输出提示信息，输入y并回车继续安装
<img src="../install/linux/conda_create.png" alt="conda_create" width="500" align="center"/>
- 激活刚创建的conda环境，在命令行中输入以下命令：
```shell
# 激活paddle_env环境
conda activate paddle_env
```
以上anaconda环境和python环境安装完毕
#### 1.3.2 Docker环境配置
**注意：第一次使用这个镜像，会自动下载该镜像，请耐心等待。您也可以访问[DockerHub](https://hub.docker.com/r/paddlepaddle/paddle/tags/)获取与您机器适配的镜像。**
```bash
# 切换到工作目录下
cd /home/Projects
# 首次运行需创建一个docker容器，再次运行时不需要运行当前命令
# 创建一个名字为ppocr的docker容器，并将当前目录映射到容器的/paddle目录下
如果您希望在CPU环境下使用docker，使用docker而不是nvidia-docker创建docker
sudo docker run --name ppocr -v $PWD:/paddle --network=host -it paddlepaddle/paddle:latest-dev-cuda10.1-cudnn7-gcc82 /bin/bash
如果使用CUDA10，请运行以下命令创建容器，设置docker容器共享内存shm-size为64G，建议设置32G以上
sudo nvidia-docker run --name ppocr -v $PWD:/paddle --shm-size=64G --network=host -it paddlepaddle/paddle:latest-dev-cuda10.1-cudnn7-gcc82 /bin/bash
# ctrl+P+Q可退出docker 容器，重新进入docker 容器使用如下命令
sudo docker container exec -it ppocr /bin/bash
```
<a name="2"></a>
## 2. 安装PaddlePaddle
- 如果您的机器安装的是CUDA9或CUDA10，请运行以下命令安装
```bash
python3 -m pip install paddlepaddle-gpu -i https://mirror.baidu.com/pypi/simple
```
- 如果您的机器是CPU，请运行以下命令安装
```bash
python3 -m pip install paddlepaddle -i https://mirror.baidu.com/pypi/simple
```
更多的版本需求，请参照[飞桨官网安装文档](https://www.paddlepaddle.org.cn/install/quick)中的说明进行操作。
doc/doc_ch/inference_ppocr.md 0 → 100644
浏览文件 @ a8a9b2e5
# PP-OCR模型库快速推理
本文介绍针对PP-OCR模型库的Python推理引擎使用方法，内容依次为文本检测、文本识别、方向分类器以及三者串联在CPU、GPU上的预测方法。
- [1. 文本检测模型推理](#文本检测模型推理)
- [2. 文本识别模型推理](#文本识别模型推理)
- [2.1 超轻量中文识别模型推理](#超轻量中文识别模型推理)
- [2.2 多语言模型的推理](#多语言模型的推理)
- [3. 方向分类模型推理](#方向分类模型推理)
- [4. 文本检测、方向分类和文字识别串联推理](#文本检测、方向分类和文字识别串联推理)
<a name="文本检测模型推理"></a>
## 1. 文本检测模型推理
文本检测模型推理，默认使用DB模型的配置参数。超轻量中文检测模型推理，可以执行如下命令：
```
# 下载超轻量中文检测模型：
wget https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_det_infer.tartar xf ch_ppocr_mobile_v2.0_det_infer.tarpython3 tools/infer/predict_det.py --image_dir="./doc/imgs/00018069.jpg" --det_model_dir="./ch_ppocr_mobile_v2.0_det_infer/"
```
可视化文本检测结果默认保存到`./inference_results`文件夹里面，结果文件的名称前缀为'det_res'。结果示例如下：
![](/Users/zhulingfeng01/OCR/PaddleOCR/doc/imgs_results/det_res_00018069.jpg)
通过参数`limit_type`和`det_limit_side_len`来对图片的尺寸进行限制，
`limit_type`可选参数为[`max`, `min`]，
`det_limit_size_len` 为正整数，一般设置为32 的倍数，比如960。
参数默认设置为`limit_type='max', det_limit_side_len=960`。表示网络输入图像的最长边不能超过960，
如果超过这个值，会对图像做等宽比的resize操作，确保最长边为`det_limit_side_len`。
设置为`limit_type='min', det_limit_side_len=960` 则表示限制图像的最短边为960。
如果输入图片的分辨率比较大，而且想使用更大的分辨率预测，可以设置det_limit_side_len 为想要的值，比如1216：
```
python3 tools/infer/predict_det.py --image_dir="./doc/imgs/1.jpg" --det_model_dir="./inference/det_db/" --det_limit_type=max --det_limit_side_len=1216
```
如果想使用CPU进行预测，执行命令如下
```
python3 tools/infer/predict_det.py --image_dir="./doc/imgs/1.jpg" --det_model_dir="./inference/det_db/" --use_gpu=False
```
<a name="文本识别模型推理"></a>
## 2. 文本识别模型推理
<a name="超轻量中文识别模型推理"></a>
### 2.1 超轻量中文识别模型推理
超轻量中文识别模型推理，可以执行如下命令：
```
# 下载超轻量中文识别模型：
wget https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_infer.tar
tar xf ch_ppocr_mobile_v2.0_rec_infer.tar
python3 tools/infer/predict_rec.py --image_dir="./doc/imgs_words/ch/word_4.jpg" --rec_model_dir="ch_ppocr_mobile_v2.0_rec_infer"
```
![](/Users/zhulingfeng01/OCR/PaddleOCR/doc/imgs_words/ch/word_4.jpg)
执行命令后，上面图像的预测结果（识别的文本和得分）会打印到屏幕上，示例如下：
```bash
Predicts of ./doc/imgs_words/ch/word_4.jpg:('实力活力', 0.98458153)
```
<a name="多语言模型的推理"></a>
### 2.2 多语言模型的推理
如果您需要预测的是其他语言模型，在使用inference模型预测时，需要通过`--rec_char_dict_path`指定使用的字典路径, 同时为了得到正确的可视化结果，
需要通过 `--vis_font_path` 指定可视化的字体路径，`doc/fonts/` 路径下有默认提供的小语种字体，例如韩文识别：
```
python3 tools/infer/predict_rec.py --image_dir="./doc/imgs_words/korean/1.jpg" --rec_model_dir="./your inference model" --rec_char_type="korean" --rec_char_dict_path="ppocr/utils/dict/korean_dict.txt" --vis_font_path="doc/fonts/korean.ttf"
```
![](/Users/zhulingfeng01/OCR/PaddleOCR/doc/imgs_words/korean/1.jpg)
执行命令后，上图的预测结果为：
``` text
Predicts of ./doc/imgs_words/korean/1.jpg:('바탕으로', 0.9948904)
```
<a name="方向分类模型推理"></a>
## 3. 方向分类模型推理
方向分类模型推理，可以执行如下命令：
```
# 下载超轻量中文方向分类器模型：
wget https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_infer.tar
tar xf ch_ppocr_mobile_v2.0_cls_infer.tar
python3 tools/infer/predict_cls.py --image_dir="./doc/imgs_words/ch/word_4.jpg" --cls_model_dir="ch_ppocr_mobile_v2.0_cls_infer"
```
![](/Users/zhulingfeng01/OCR/PaddleOCR/doc/imgs_words/ch/word_1.jpg)
执行命令后，上面图像的预测结果（分类的方向和得分）会打印到屏幕上，示例如下：
```
Predicts of ./doc/imgs_words/ch/word_4.jpg:['0', 0.9999982]
```
<a name="文本检测、方向分类和文字识别串联推理"></a>
## 4. 文本检测、方向分类和文字识别串联推理
以超轻量中文OCR模型推理为例，在执行预测时，需要通过参数`image_dir`指定单张图像或者图像集合的路径、参数`det_model_dir`,`cls_model_dir`和`rec_model_dir`分别指定检测，方向分类和识别的inference模型路径。参数`use_angle_cls`用于控制是否启用方向分类模型。`use_mp`表示是否使用多进程。`total_process_num`表示在使用多进程时的进程数。可视化识别结果默认保存到 ./inference_results 文件夹里面。
```shell
# 使用方向分类器
python3 tools/infer/predict_system.py --image_dir="./doc/imgs/00018069.jpg" --det_model_dir="./inference/det_db/" --cls_model_dir="./inference/cls/" --rec_model_dir="./inference/rec_crnn/" --use_angle_cls=true
# 不使用方向分类器
python3 tools/infer/predict_system.py --image_dir="./doc/imgs/00018069.jpg" --det_model_dir="./inference/det_db/" --rec_model_dir="./inference/rec_crnn/" --use_angle_cls=false
# 使用多进程
python3 tools/infer/predict_system.py --image_dir="./doc/imgs/00018069.jpg" --det_model_dir="./inference/det_db/" --rec_model_dir="./inference/rec_crnn/" --use_angle_cls=false --use_mp=True --total_process_num=6
```
执行命令后，识别结果图像如下：
![](/Users/zhulingfeng01/OCR/PaddleOCR/doc/imgs_results/system_res_00018069.jpg)
doc/doc_ch/knowledge_distillation.md
浏览文件 @ a8a9b2e5
......@@ -39,7 +39,7 @@ PaddleOCR中集成了知识蒸馏的算法，具体地，有以下几个主要
### 2.1 识别配置文件解析
配置文件在[rec_chinese_lite_train_distillation_v2.1.yml](../../configs/rec/ch_ppocr_v2.1/rec_chinese_lite_train_distillation_v2.1.yml)。
配置文件在[ch_PP-OCRv2_rec.yml](../../configs/rec/ch_PP-OCRv2/ch_PP-OCRv2_rec.yml)。
#### 2.1.1 模型结构
......
doc/doc_ch/models_and_config.md 0 → 100644
浏览文件 @ a8a9b2e5
# PP-OCR模型与配置文件
PP-OCR模型与配置文件一章主要补充一些OCR模型的基本概念、配置文件的内容与作用以便对模型后续的参数调整和训练中拥有更好的体验。
本章包含三个部分，首先在[PP-OCR模型下载](./models_list.md)中解释PP-OCR模型的类型概念，并提供所有模型的下载链接。然后在[配置文件内容与生成](./config.md)中详细说明调整PP-OCR模型所需的参数。最后的[模型库快速使用](./inference_ppocr.md)是对第一节PP-OCR模型库使用方法的介绍，可以通过Python推理引擎快速利用丰富的模型库模型获得测试结果。
------
下面我们首先了解一些OCR相关的基本概念：
- [1. OCR 简要介绍](#1-ocr-----)
* [1.1 OCR 检测模型基本概念](#11-ocr---------)
* [1.2 OCR 识别模型基本概念](#12-ocr---------)
* [1.3 PP-OCR模型](#13-pp-ocr--)
<a name="1-ocr-----"></a>
## 1. OCR 简要介绍
本节简要介绍OCR检测模型、识别模型的基本概念，并介绍PaddleOCR的PP-OCR模型。
OCR（Optical Character Recognition，光学字符识别）目前是文字识别的统称，已不限于文档或书本文字识别，更包括识别自然场景下的文字，又可以称为STR（Scene Text Recognition）。
OCR文字识别一般包括两个部分，文本检测和文本识别；文本检测首先利用检测算法检测到图像中的文本行；然后检测到的文本行用识别算法去识别到具体文字。
<a name="11-ocr---------"></a>
### 1.1 OCR 检测模型基本概念
文本检测就是要定位图像中的文字区域，然后通常以边界框的形式将单词或文本行标记出来。传统的文字检测算法多是通过手工提取特征的方式，特点是速度快，简单场景效果好，但是面对自然场景，效果会大打折扣。当前多是采用深度学习方法来做。
基于深度学习的文本检测算法可以大致分为以下几类：
1. 基于目标检测的方法；一般是预测得到文本框后，通过NMS筛选得到最终文本框，多是四点文本框，对弯曲文本场景效果不理想。典型算法为EAST、Text Box等方法。
2. 基于分割的方法；将文本行当成分割目标，然后通过分割结果构建外接文本框，可以处理弯曲文本，对于文本交叉场景问题效果不理想。典型算法为DB、PSENet等方法。
3. 混合目标检测和分割的方法；
<a name="12-ocr---------"></a>
### 1.2 OCR 识别模型基本概念
OCR识别算法的输入数据一般是文本行，背景信息不多，文字占据主要部分，识别算法目前可以分为两类算法：
1. 基于CTC的方法；即识别算法的文字预测模块是基于CTC的，常用的算法组合为CNN+RNN+CTC。目前也有一些算法尝试在网络中加入transformer模块等等。
2. 基于Attention的方法；即识别算法的文字预测模块是基于Attention的，常用算法组合是CNN+RNN+Attention。
<a name="13-pp-ocr--"></a>
### 1.3 PP-OCR模型
PaddleOCR 中集成了很多OCR算法，文本检测算法有DB、EAST、SAST等等，文本识别算法有CRNN、RARE、StarNet、Rosetta、SRN等算法。
其中PaddleOCR针对中英文自然场景通用OCR，推出了PP-OCR系列模型，PP-OCR模型由DB+CRNN算法组成，利用海量中文数据训练加上模型调优方法，在中文场景上具备较高的文本检测识别能力。并且PaddleOCR推出了高精度超轻量PP-OCRv2模型，检测模型仅3M，识别模型仅8.5M，利用[PaddleSlim](https://github.com/PaddlePaddle/PaddleSlim)的模型量化方法，可以在保持精度不降低的情况下，将检测模型压缩到0.8M，识别压缩到3M，更加适用于移动端部署场景。
doc/doc_ch/models_list.md
浏览文件 @ a8a9b2e5
## OCR模型列表（V2.0，2021年1月20日更新）
## OCR模型列表（V2.1，2021年9月6日更新）
> **说明**
> 1. 2.0版模型和[1.1版模型](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/doc/doc_ch/models_list.md) 的主要区别在于动态图训练vs.静态图训练，模型性能上无明显差距。
> 2. 本文档提供的是PPOCR自研模型列表，更多基于公开数据集的算法介绍与预训练模型可以参考：[算法概览文档](./algorithm_overview.md)。
> 1. 2.1版模型相比2.0版模型，2.1的模型在模型精度上做了提升
> 2. 2.0版模型和[1.1版模型](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/doc/doc_ch/models_list.md) 的主要区别在于动态图训练vs.静态图训练，模型性能上无明显差距。
> 3. 本文档提供的是PPOCR自研模型列表，更多基于公开数据集的算法介绍与预训练模型可以参考：[算法概览文档](./algorithm_overview.md)。
- [一、文本检测模型](#文本检测模型)
......@@ -32,6 +33,8 @@ PaddleOCR提供的可下载模型包括`推理模型`、`训练模型`、`预训
|模型名称|模型简介|配置文件|推理模型大小|下载地址|
| --- | --- | --- | --- | --- |
|ch_PP-OCRv2_det_slim|slim量化+蒸馏版超轻量模型，支持中英文、多语种文本检测|[ch_PP-OCRv2_det_cml.yml](../../configs/det/ch_PP-OCRv2/ch_PP-OCR_det_cml.yml)| 3M |[推理模型](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_det_slim_quant_infer.tar)|
|ch_PP-OCRv2_det|原始超轻量模型，支持中英文、多语种文本检测|[ch_PP-OCRv2_det_cml.yml](../../configs/det/ch_PP-OCRv2/ch_PP-OCR_det_cml.yml)|3M|[推理模型](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_det_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_det_distill_train.tar)|
|ch_ppocr_mobile_slim_v2.0_det|slim裁剪版超轻量模型，支持中英文、多语种文本检测|[ch_det_mv3_db_v2.0.yml](../../configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml)| 2.6M |[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/slim/ch_ppocr_mobile_v2.0_det_prune_infer.tar)|
|ch_ppocr_mobile_v2.0_det|原始超轻量模型，支持中英文、多语种文本检测|[ch_det_mv3_db_v2.0.yml](../../configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml)|3M|[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_det_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_det_train.tar)|
|ch_ppocr_server_v2.0_det|通用模型，支持中英文、多语种文本检测，比超轻量模型更大，但效果更好|[ch_det_res18_db_v2.0.yml](../../configs/det/ch_ppocr_v2.0/ch_det_res18_db_v2.0.yml)|47M|[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_det_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_det_train.tar)|
......@@ -45,6 +48,8 @@ PaddleOCR提供的可下载模型包括`推理模型`、`训练模型`、`预训
|模型名称|模型简介|配置文件|推理模型大小|下载地址|
| --- | --- | --- | --- | --- |
|ch_PP-OCRv2_rec_slim|slim量化版超轻量模型，支持中英文、数字识别|[ch_PP-OCRv2_rec.yml](../../configs/rec/ch_PP-OCRv2/ch_PP-OCRv2_rec.yml)| 9M |[推理模型](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_rec_slim_quant_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_rec_slim_quant_train.tar) |
|ch_PP-OCRv2_rec|原始超轻量模型，支持中英文、数字识别|[ch_PP-OCRv2_rec.yml](../../configs/rec/ch_PP-OCRv2/ch_PP-OCRv2_rec.yml)|8.5M|[推理模型](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_rec_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_rec_train.tar) |
|ch_ppocr_mobile_slim_v2.0_rec|slim裁剪量化版超轻量模型，支持中英文、数字识别|[rec_chinese_lite_train_v2.0.yml](../../configs/rec/ch_ppocr_v2.0/rec_chinese_lite_train_v2.0.yml)| 6M |[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_slim_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_slim_train.tar) |
|ch_ppocr_mobile_v2.0_rec|原始超轻量模型，支持中英文、数字识别|[rec_chinese_lite_train_v2.0.yml](../../configs/rec/ch_ppocr_v2.0/rec_chinese_lite_train_v2.0.yml)|5.2M|[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_train.tar) / [预训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_pre.tar) |
|ch_ppocr_server_v2.0_rec|通用模型，支持中英文、数字识别|[rec_chinese_common_train_v2.0.yml](../../configs/rec/ch_ppocr_v2.0/rec_chinese_common_train_v2.0.yml)|94.8M|[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_rec_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_rec_train.tar) / [预训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_rec_pre.tar) |
......@@ -62,46 +67,6 @@ PaddleOCR提供的可下载模型包括`推理模型`、`训练模型`、`预训
<a name="多语言识别模型"></a>
#### 3. 多语言识别模型（更多语言持续更新中...）
**说明：** 新增的多语言模型的配置文件通过代码方式生成，您可以通过`--help`参数查看当前PaddleOCR支持生成哪些多语言的配置文件：
```bash
# 该代码需要在指定目录运行
cd {your/path/}PaddleOCR/configs/rec/multi_language/
python3 generate_multi_language_configs.py --help
```
下面以生成意大利语配置文件为例：
##### 1. 生成意大利语配置文件测试现有模型
如果您仅仅想用配置文件测试PaddleOCR提供的多语言模型可以通过下面命令生成默认的配置文件，使用PaddleOCR提供的小语种字典进行预测。
```bash
# 该代码需要在指定目录运行
cd {your/path/}PaddleOCR/configs/rec/multi_language/
# 通过-l或者--language参数设置需要生成的语种的配置文件，该命令会将默认参数写入配置文件
python3 generate_multi_language_configs.py -l it
```
##### 2. 生成意大利语配置文件训练自己的数据
如果您想训练自己的小语种模型，可以准备好训练集文件、验证集文件、字典文件和训练数据路径，这里假设准备的意大利语的训练集、验证集、字典和训练数据路径为：
- 训练集:{your/path/}PaddleOCR/train_data/train_list.txt
- 验证集:{your/path/}PaddleOCR/train_data/val_list.txt
- 使用PaddleOCR提供的默认字典：{your/path/}PaddleOCR/ppocr/utils/dict/it_dict.txt
- 训练数据路径:{your/path/}PaddleOCR/train_data
使用以下命令生成配置文件：
```bash
# 该代码需要在指定目录运行
cd {your/path/}PaddleOCR/configs/rec/multi_language/
# -l或者--language字段是必须的
# --train修改训练集，--val修改验证集，--data_dir修改数据集目录，-o修改对应默认参数
# --dict命令改变字典路径，示例使用默认字典路径则该参数可不填
python3 generate_multi_language_configs.py -l it \
--train train_data/train_list.txt \
--val train_data/val_list.txt \
--data_dir train_data \
-o Global.use_gpu=False
```
<a name="多语言模型与配置文件"></a>
##### 3. 多语言模型与配置文件
|模型名称|字典文件|模型简介|配置文件|推理模型大小|下载地址|
| --- | --- | --- | --- |--- | --- |
| french_mobile_v2.0_rec | ppocr/utils/dict/french_dict.txt |法文识别|[rec_french_lite_train.yml](../../configs/rec/multi_language/rec_french_lite_train.yml)|2.65M|[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/french_mobile_v2.0_rec_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/french_mobile_v2.0_rec_train.tar) |
......@@ -125,13 +90,16 @@ python3 generate_multi_language_configs.py -l it \
|模型名称|模型简介|配置文件|推理模型大小|下载地址|
| --- | --- | --- | --- | --- |
|ch_ppocr_mobile_slim_v2.0_cls|slim量化版模型|[cls_mv3.yml](../../configs/cls/cls_mv3.yml)| 2.1M |[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_slim_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_slim_infer.tar) |
|ch_ppocr_mobile_v2.0_cls|原始模型|[cls_mv3.yml](../../configs/cls/cls_mv3.yml)|1.38M|[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_train.tar) |
|ch_ppocr_mobile_slim_v2.0_cls|slim量化版模型，对检测到的文本行文字角度分类|[cls_mv3.yml](../../configs/cls/cls_mv3.yml)| 2.1M |[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_slim_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_slim_infer.tar) |
|ch_ppocr_mobile_v2.0_cls|原始分类器模型，对检测到的文本行文字角度分类|[cls_mv3.yml](../../configs/cls/cls_mv3.yml)|1.38M|[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_train.tar) |
<a name="Paddle-Lite模型"></a>
### 四、Paddle-Lite 模型
|模型版本|模型简介|模型大小|检测模型|文本方向分类模型|识别模型|Paddle-Lite版本|
|---|---|---|---|---|---|---|
|V2.0|超轻量中文OCR 移动端模型|7.8M|[下载地址](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_det_opt.nb)|[下载地址](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_cls_opt.nb)|[下载地址](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_rec_opt.nb)|v2.9|
|V2.0(slim)|超轻量中文OCR 移动端模型|3.3M|[下载地址](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_det_slim_opt.nb)|[下载地址](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_cls_slim_opt.nb)|[下载地址](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_rec_slim_opt.nb)|v2.9|
|PP-OCRv2|蒸馏版超轻量中文OCR移动端模型|11M|[下载地址](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_det_infer_opt.nb)|[下载地址](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_cls_opt.nb)|[下载地址](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_rec_infer_opt.nb)|v2.9|
|PP-OCRv2(slim)|蒸馏版超轻量中文OCR移动端模型|4.9M|[下载地址](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_det_slim_opt.nb)|[下载地址](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_cls_slim_opt.nb)|[下载地址](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_rec_slim_opt.nb)|v2.9|
|V2.0|ppocr_v2.0超轻量中文OCR移动端模型|7.8M|[下载地址](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_det_opt.nb)|[下载地址](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_cls_opt.nb)|[下载地址](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_rec_opt.nb)|v2.9|
|V2.0(slim)|ppocr_v2.0超轻量中文OCR移动端模型|3.3M|[下载地址](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_det_slim_opt.nb)|[下载地址](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_cls_slim_opt.nb)|[下载地址](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_rec_slim_opt.nb)|v2.9|
doc/doc_ch/multi_languages.md
浏览文件 @ a8a9b2e5
......@@ -200,9 +200,9 @@ ppocr 支持使用自己的数据进行自定义训练或finetune, 其中识别
|英文|english|en| |乌克兰文|Ukranian|uk|
|法文|french|fr| |白俄罗斯文|Belarusian|be|
|德文|german|german| |泰卢固文|Telugu |te|
|日文|japan|japan| | |阿巴扎文|Abaza |abq|
|日文|japan|japan| | 阿巴扎文 | Abaza | abq |
|韩文|korean|korean| |泰米尔文|Tamil |ta|
|中文繁体|chinese traditional |ch_tra| |南非荷兰文 |Afrikaans |af|
|中文繁体|chinese traditional |chinese_cht| |南非荷兰文 |Afrikaans |af|
|意大利文| Italian |it| |阿塞拜疆文 |Azerbaijani |az|
|西班牙文|Spanish |es| |波斯尼亚文|Bosnian|bs|
|葡萄牙文| Portuguese|pt| |捷克文|Czech|cs|
......
doc/doc_ch/paddleOCR_overview.md 0 → 100644
浏览文件 @ a8a9b2e5
# PaddleOCR全景图与项目克隆
## 1. PaddleOCR全景图
PaddleOCR包含丰富的文本检测、文本识别以及端到端算法。结合实际测试与产业经验，PaddleOCR选择DB和CRNN作为基础的检测和识别模型，经过一系列优化策略提出面向产业应用的PP-OCR模型。PP-OCR模型针对通用场景，根据不同语种形成了PP-OCR模型库。基于PP-OCR的能力，PaddleOCR针对文档场景任务发布PP-Structure工具库，包含版面分析和表格识别两大任务。为了打通产业落地的全流程，PaddleOCR提供了规模化的数据生产工具和多种预测部署工具，助力开发者快速落地。
<div align="center">
<img src="../overview.png">
</div>
## 2. 项目克隆
### **2.1 克隆PaddleOCR repo代码**
```
【推荐】git clone https://github.com/PaddlePaddle/PaddleOCR
```
如果因为网络问题无法pull成功，也可选择使用码云上的托管：
```
git clone https://gitee.com/paddlepaddle/PaddleOCR
```
注：码云托管代码可能无法实时同步本github项目更新，存在3~5天延时，请优先使用推荐方式。
### **2.2 安装第三方库**
```
cd PaddleOCR
pip3 install -r requirements.txt
```
doc/doc_ch/quickstart.md
浏览文件 @ a8a9b2e5
# PaddleOCR快速开始
# 中文OCR模型快速使用
## 1.环境配置
- [PaddleOCR快速开始](#paddleocr)
请先参考[快速安装](./installation.md)配置PaddleOCR运行环境。
+ [1. 安装PaddleOCR whl包](#1)
* [2. 便捷使用](#2)
+ [2.1 命令行使用](#21)
- [2.1.1 中英文模型](#211)
- [2.1.2 多语言模型](#212)
- [2.1.3 版面分析](#213)
+ [2.2 Python脚本使用](#22)
- [2.2.1 中英文与多语言使用](#221)
- [2.2.2 版面分析](#222)
*注意：也可以通过 whl 包安装使用PaddleOCR，具体参考[Paddleocr Package使用说明](./whl.md)。*
<a name="1"></a>
## 2.inference模型下载
## 1. 安装PaddleOCR whl包
* 移动端和服务器端的检测与识别模型如下，更多模型下载（包括多语言），可以参考[PP-OCR v2.0 系列模型下载](../doc_ch/models_list.md)
```bash
pip install "paddleocr>=2.0.1" # 推荐使用2.0.1+版本
```
| 模型简介 | 模型名称 |推荐场景 | 检测模型 | 方向分类器 | 识别模型 |
| ------------ | --------------- | ----------------|---- | ---------- | -------- |
| 中英文超轻量OCR模型（8.1M） | ch_ppocr_mobile_v2.0_xx |移动端&服务器端|[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_det_infer.tar) / [预训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_det_train.tar)|[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_infer.tar) / [预训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_train.tar) |[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_infer.tar) / [预训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_pre.tar) |
| 中英文通用OCR模型（143M） | ch_ppocr_server_v2.0_xx |服务器端 |[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_det_infer.tar) / [预训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_det_train.tar) |[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_infer.tar) / [预训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_train.tar) |[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_rec_infer.tar) / [预训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_rec_pre.tar) |
- 对于Windows环境用户：
直接通过pip安装的shapely库可能出现`[winRrror 126] 找不到指定模块的问题`。建议从[这里](https://www.lfd.uci.edu/~gohlke/pythonlibs/#shapely)下载shapely安装包完成安装，
* windows 环境下如果没有安装wget,下载模型时可将链接复制到浏览器中下载，并解压放置在相应目录下
- 使用**版面分析**功能时，运行以下命令**安装 Layout-Parser**
复制上表中的检测和识别的`inference模型`下载地址，并解压
```bash
pip3 install -U https://paddleocr.bj.bcebos.com/whl/layoutparser-0.0.0-py3-none-any.whl
```
```
mkdir inference && cd inference
# 下载检测模型并解压
wget {url/of/detection/inference_model} && tar xf {name/of/detection/inference_model/package}
# 下载识别模型并解压
wget {url/of/recognition/inference_model} && tar xf {name/of/recognition/inference_model/package}
# 下载方向分类器模型并解压
wget {url/of/classification/inference_model} && tar xf {name/of/classification/inference_model/package}
cd ..
```
以超轻量级模型为例：
<a name="2"></a>
## 2. 便捷使用
<a name="21"></a>
### 2.1 命令行使用
PaddleOCR提供了一系列测试图片，点击[这里](https://paddleocr.bj.bcebos.com/dygraph_v2.1/ppocr_img.zip)下载并解压，然后在终端中切换到相应目录
```
mkdir inference && cd inference
# 下载超轻量级中文OCR模型的检测模型并解压
wget https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_det_infer.tar && tar xf ch_ppocr_mobile_v2.0_det_infer.tar
# 下载超轻量级中文OCR模型的识别模型并解压
wget https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_infer.tar && tar xf ch_ppocr_mobile_v2.0_rec_infer.tar
# 下载超轻量级中文OCR模型的文本方向分类器模型并解压
wget https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_infer.tar && tar xf ch_ppocr_mobile_v2.0_cls_infer.tar
cd ..
cd /path/to/ppocr_img
```
解压完毕后应有如下文件结构：
如果不使用提供的测试图片，可以将下方`--image_dir`参数替换为相应的测试图片路径
<a name="211"></a>
#### 2.1.1 中英文模型
* 检测+方向分类器+识别全流程：设置方向分类器参数`--use_angle_cls true`后可对竖排文本进行识别。
```bash
paddleocr --image_dir ./imgs/11.jpg --use_angle_cls true
```
结果是一个list，每个item包含了文本框，文字和识别置信度
```bash
[[[24.0, 36.0], [304.0, 34.0], [304.0, 72.0], [24.0, 74.0]], ['纯臻营养护发素', 0.964739]]
[[[24.0, 80.0], [172.0, 80.0], [172.0, 104.0], [24.0, 104.0]], ['产品信息/参数', 0.98069626]]
[[[24.0, 109.0], [333.0, 109.0], [333.0, 136.0], [24.0, 136.0]], ['（45元/每公斤，100公斤起订）', 0.9676722]]
......
```
- 单独使用检测：设置`--rec`为`false`
```bash
paddleocr --image_dir ./imgs/11.jpg --rec false
```
结果是一个list，每个item只包含文本框
```bash
[[26.0, 457.0], [137.0, 457.0], [137.0, 477.0], [26.0, 477.0]]
[[25.0, 425.0], [372.0, 425.0], [372.0, 448.0], [25.0, 448.0]]
[[128.0, 397.0], [273.0, 397.0], [273.0, 414.0], [128.0, 414.0]]
......
```
- 单独使用识别：设置`--det`为`false`
```bash
paddleocr --image_dir ./imgs_words/ch/word_1.jpg --det false
```
结果是一个list，每个item只包含识别结果和识别置信度
```bash
['韩国小馆', 0.9907421]
```
如需使用2.0模型，请指定参数`--version 2.0`，paddleocr默认使用2.1模型。更多whl包使用可参考[whl包文档](./whl.md)
<a name="212"></a>
#### 2.1.2 多语言模型
Paddleocr目前支持80个语种，可以通过修改`--lang`参数进行切换，对于英文模型，指定`--lang=en`。
``` bash
paddleocr --image_dir ./imgs_en/254.jpg --lang=en
```
├── ch_ppocr_mobile_v2.0_cls_infer
│ ├── inference.pdiparams
│ ├── inference.pdiparams.info
│ └── inference.pdmodel
├── ch_ppocr_mobile_v2.0_det_infer
│ ├── inference.pdiparams
│ ├── inference.pdiparams.info
│ └── inference.pdmodel
├── ch_ppocr_mobile_v2.0_rec_infer
├── inference.pdiparams
├── inference.pdiparams.info
└── inference.pdmodel
<div align="center">
<img src="../imgs_en/254.jpg" width="300" height="600">
<img src="../imgs_results/multi_lang/img_02.jpg" width="600" height="600">
</div>
结果是一个list，每个item包含了文本框，文字和识别置信度
```text
[('PHO CAPITAL', 0.95723116), [[66.0, 50.0], [327.0, 44.0], [327.0, 76.0], [67.0, 82.0]]]
[('107 State Street', 0.96311164), [[72.0, 90.0], [451.0, 84.0], [452.0, 116.0], [73.0, 121.0]]]
[('Montpelier Vermont', 0.97389287), [[69.0, 132.0], [501.0, 126.0], [501.0, 158.0], [70.0, 164.0]]]
[('8022256183', 0.99810505), [[71.0, 175.0], [363.0, 170.0], [364.0, 202.0], [72.0, 207.0]]]
[('REG 07-24-201706:59 PM', 0.93537045), [[73.0, 299.0], [653.0, 281.0], [654.0, 318.0], [74.0, 336.0]]]
[('045555', 0.99346405), [[509.0, 331.0], [651.0, 325.0], [652.0, 356.0], [511.0, 362.0]]]
[('CT1', 0.9988654), [[535.0, 367.0], [654.0, 367.0], [654.0, 406.0], [535.0, 406.0]]]
......
```
## 3.单张图像或者图像集合预测
常用的多语言简写包括
以下代码实现了文本检测、方向分类器和识别串联推理，在执行预测时，需要通过参数image_dir指定单张图像或者图像集合的路径、参数`det_model_dir`指定检测inference模型的路径、参数`rec_model_dir`指定识别inference模型的路径、参数`use_angle_cls`指定是否使用方向分类器、参数`cls_model_dir`指定方向分类器inference模型的路径、参数`use_space_char`指定是否预测空格字符。可视化识别结果默认保存到`./inference_results`文件夹里面。
| 语种 | 缩写 | | 语种 | 缩写 | | 语种 | 缩写 |
| -------- | ----------- | ---- | -------- | ------ | ---- | -------- | ------ |
| 中文 | ch | | 法文 | fr | | 日文 | japan |
| 英文 | en | | 德文 | german | | 韩文 | korean |
| 繁体中文 | chinese_cht | | 意大利文 | it | | 俄罗斯文 | ru |
```bash
全部语种及其对应的缩写列表可查看[多语言模型教程](./multi_languages.md)
<a name="213"></a>
#### 2.1.3 版面分析
# 预测image_dir指定的单张图像
python3 tools/infer/predict_system.py --image_dir="./doc/imgs/11.jpg" --det_model_dir="./inference/ch_ppocr_mobile_v2.0_det_infer/" --rec_model_dir="./inference/ch_ppocr_mobile_v2.0_rec_infer/" --cls_model_dir="./inference/ch_ppocr_mobile_v2.0_cls_infer/" --use_angle_cls=True --use_space_char=True
版面分析是指对文档图片中的文字、标题、列表、图片和表格5类区域进行划分。对于前三类区域，直接使用OCR模型完成对应区域文字检测与识别，并将结果保存在txt中。对于表格类区域，经过表格结构化处理后，表格图片转换为相同表格样式的Excel文件。图片区域会被单独裁剪成图像。
# 预测image_dir指定的图像集合
python3 tools/infer/predict_system.py --image_dir="./doc/imgs/" --det_model_dir="./inference/ch_ppocr_mobile_v2.0_det_infer/" --rec_model_dir="./inference/ch_ppocr_mobile_v2.0_rec_infer/" --cls_model_dir="./inference/ch_ppocr_mobile_v2.0_cls_infer/" --use_angle_cls=True --use_space_char=True
使用PaddleOCR的版面分析功能，需要指定`--type=structure`
# 如果想使用CPU进行预测，需设置use_gpu参数为False
python3 tools/infer/predict_system.py --image_dir="./doc/imgs/11.jpg" --det_model_dir="./inference/ch_ppocr_mobile_v2.0_det_infer/" --rec_model_dir="./inference/ch_ppocr_mobile_v2.0_rec_infer/" --cls_model_dir="./inference/ch_ppocr_mobile_v2.0_cls_infer/" --use_angle_cls=True --use_space_char=True --use_gpu=False
```bash
paddleocr --image_dir=./table/1.png --type=structure
```
- 通用中文OCR模型
- **返回结果说明**
PP-Structure的返回结果为一个dict组成的list，示例如下
```shell
[{ 'type': 'Text',
'bbox': [34, 432, 345, 462],
'res': ([[36.0, 437.0, 341.0, 437.0, 341.0, 446.0, 36.0, 447.0], [41.0, 454.0, 125.0, 453.0, 125.0, 459.0, 41.0, 460.0]],
[('Tigure-6. The performance of CNN and IPT models using difforen', 0.90060663), ('Tent ', 0.465441)])
}
]
```
其中各个字段说明如下
| 字段 | 说明 |
| ---- | ------------------------------------------------------------ |
| type | 图片区域的类型 |
| bbox | 图片区域的在原图的坐标，分别[左上角x，左上角y，右下角x，右下角y] |
| res | 图片区域的OCR或表格识别结果。<br>表格: 表格的HTML字符串; <br>OCR: 一个包含各个单行文字的检测坐标和识别结果的元组 |
运行完成后，每张图片会在`output`字段指定的目录下有一个同名目录，图片里的每个表格会存储为一个excel，图片区域会被裁剪之后保存下来，excel文件和图片名为表格在图片里的坐标。
```
/output/table/1/
└─ res.txt
└─ [454, 360, 824, 658].xlsx 表格识别结果
└─ [16, 2, 828, 305].jpg 被裁剪出的图片区域
└─ [17, 361, 404, 711].xlsx 表格识别结果
```
- **参数说明**
| 字段 | 说明 | 默认值 |
| --------------- | ---------------------------------------- | -------------------------------------------- |
| output | excel和识别结果保存的地址 | ./output/table |
| table_max_len | 表格结构模型预测时，图像的长边resize尺度 | 488 |
| table_model_dir | 表格结构模型 inference 模型地址 | None |
| table_char_type | 表格结构模型所用字典地址 | ../ppocr/utils/dict/table_structure_dict.txt |
大部分参数和paddleocr whl包保持一致，见 [whl包文档](./whl.md)
<a name="22"></a>
### 2.2 Python脚本使用
<a name="221"></a>
#### 2.2.1 中英文与多语言使用
请按照上述步骤下载相应的模型，并且更新相关的参数，示例如下：
通过Python脚本使用PaddleOCR whl包，whl包会自动下载ppocr轻量级模型作为默认模型。
* 检测+方向分类器+识别全流程
```python
from paddleocr import PaddleOCR, draw_ocr
# Paddleocr目前支持的多语言语种可以通过修改lang参数进行切换
# 例如`ch`, `en`, `fr`, `german`, `korean`, `japan`
ocr = PaddleOCR(use_angle_cls=True, lang="ch") # need to run only once to download and load model into memory
img_path = './imgs/11.jpg'
result = ocr.ocr(img_path, cls=True)
for line in result:
print(line)
# 显示结果
from PIL import Image
image = Image.open(img_path).convert('RGB')
boxes = [line[0] for line in result]
txts = [line[1][0] for line in result]
scores = [line[1][1] for line in result]
im_show = draw_ocr(image, boxes, txts, scores, font_path='./fonts/simfang.ttf')
im_show = Image.fromarray(im_show)
im_show.save('result.jpg')
```
结果是一个list，每个item包含了文本框，文字和识别置信度
```bash
# 预测image_dir指定的单张图像
python3 tools/infer/predict_system.py --image_dir="./doc/imgs/11.jpg" --det_model_dir="./inference/ch_ppocr_server_v2.0_det_infer/" --rec_model_dir="./inference/ch_ppocr_server_v2.0_rec_infer/" --cls_model_dir="./inference/ch_ppocr_mobile_v2.0_cls_infer/" --use_angle_cls=True --use_space_char=True
[[[24.0, 36.0], [304.0, 34.0], [304.0, 72.0], [24.0, 74.0]], ['纯臻营养护发素', 0.964739]]
[[[24.0, 80.0], [172.0, 80.0], [172.0, 104.0], [24.0, 104.0]], ['产品信息/参数', 0.98069626]]
[[[24.0, 109.0], [333.0, 109.0], [333.0, 136.0], [24.0, 136.0]], ['（45元/每公斤，100公斤起订）', 0.9676722]]
......
```
* 注意：
- 如果希望使用不支持空格的识别模型，在预测的时候需要注意：请将代码更新到最新版本，并添加参数 `--use_space_char=False`。
- 如果不希望使用方向分类器，在预测的时候需要注意：请将代码更新到最新版本，并添加参数 `--use_angle_cls=False`。
结果可视化
<div align="center">
<img src="../imgs_results/whl/11_det_rec.jpg" width="800">
</div>
<a name="222"></a>
更多的文本检测、识别串联推理使用方式请参考文档教程中[基于Python预测引擎推理](./inference.md)。
#### 2.2.2 版面分析
此外，文档教程中也提供了中文OCR模型的其他预测部署方式：
- [基于C++预测引擎推理](../../deploy/cpp_infer/readme.md)
- [服务部署](../../deploy/hubserving)
- [端侧部署（目前只支持静态图）](https://github.com/PaddlePaddle/PaddleOCR/tree/develop/deploy/lite)
```python
import os
import cv2
from paddleocr import PPStructure,draw_structure_result,save_structure_res
table_engine = PPStructure(show_log=True)
save_folder = './output/table'
img_path = './table/paper-image.jpg'
img = cv2.imread(img_path)
result = table_engine(img)
save_structure_res(result, save_folder,os.path.basename(img_path).split('.')[0])
for line in result:
line.pop('img')
print(line)
from PIL import Image
font_path = './fonts/simfang.ttf' # PaddleOCR下提供字体包
image = Image.open(img_path).convert('RGB')
im_show = draw_structure_result(image, result,font_path=font_path)
im_show = Image.fromarray(im_show)
im_show.save('result.jpg')
```
doc/doc_ch/recognition.md
浏览文件 @ a8a9b2e5
## 文字识别
# 文字识别
本文提供了PaddleOCR文本识别任务的全流程指南，包括数据准备、模型训练、调优、评估、预测，各个阶段的详细说明：
- [1 数据准备](#数据准备)
- [1.1 自定义数据集](#自定义数据集)
- [1.2 数据下载](#数据下载)
- [1.3 字典](#字典)
- [1.4 支持空格](#支持空格)
- [2 启动训练](#启动训练)
- [2.1 数据增强](#数据增强)
- [2.2 训练](#训练)
- [2.3 小语种](#小语种)
- [2.2 通用模型训练](#通用模型训练)
- [2.3 多语言模型训练](#多语言模型训练)
- [3 评估](#评估)
- [4 预测](#预测)
- [4.1 训练引擎预测](#训练引擎预测)
- [5 转Inference模型测试](#Inference)
<a name="数据准备"></a>
### 1. 数据准备
## 1. 数据准备
PaddleOCR 支持两种数据格式:
- `lmdb` 用于训练以lmdb格式存储的数据集;
- `通用数据` 用于训练以文本文件存储的数据集:
- `lmdb` 用于训练以lmdb格式存储的数据集(LMDBDataSet);
- `通用数据` 用于训练以文本文件存储的数据集(SimpleDataSet);
训练数据的默认存储路径是 `PaddleOCR/train_data`,如果您的磁盘上已有数据集，只需创建软链接至数据集目录：
......@@ -36,7 +34,7 @@ mklink /d <path/to/paddle_ocr>/train_data/dataset <path/to/dataset>
```
<a name="准备数据集"></a>
#### 1.1 自定义数据集
### 1.1 自定义数据集
下面以通用数据集为例， 介绍如何准备数据集：
* 训练集
......@@ -82,14 +80,13 @@ train_data/rec/train/word_002.jpg 用科技让复杂的世界更简单
<a name="数据下载"></a>
1.2 数据下载
若您本地没有数据集，可以在官网下载 [icdar2015](http://rrc.cvc.uab.es/?ch=4&com=downloads) 数据，用于快速验证。也可以参考[DTRB](https://github.com/clovaai/deep-text-recognition-benchmark#download-lmdb-dataset-for-traininig-and-evaluation-from-here) ，下载 benchmark 所需的lmdb格式数据集。
### 1.2 数据下载
如果你使用的是icdar2015的公开数据集，PaddleOCR 提供了一份用于训练 icdar2015 数据集的标签文件，通过以下方式下载：
- ICDAR2015
如果希望复现SRN的论文指标，需要下载离线[增广数据](https://pan.baidu.com/s/1-HSZ-ZVdqBF2HaBZ5pRAKA),提取码: y3ry。增广数据是由MJSynth和SynthText做旋转和扰动得到的。数据下载完成后请解压到 {your_path}/PaddleOCR/train_data/data_lmdb_release/training/ 路径下。
若您本地没有数据集，可以在官网下载 [ICDAR2015](http://rrc.cvc.uab.es/?ch=4&com=downloads) 数据，用于快速验证。也可以参考[DTRB](https://github.com/clovaai/deep-text-recognition-benchmark#download-lmdb-dataset-for-traininig-and-evaluation-from-here) ，下载 benchmark 所需的lmdb格式数据集。
如果你使用的是icdar2015的公开数据集，PaddleOCR 提供了一份用于训练 ICDAR2015 数据集的标签文件，通过以下方式下载：
```
# 训练集标签
wget -P ./train_data/ic15_data https://paddleocr.bj.bcebos.com/dataset/rec_gt_train.txt
......@@ -97,15 +94,25 @@ wget -P ./train_data/ic15_data https://paddleocr.bj.bcebos.com/dataset/rec_gt_t
wget -P ./train_data/ic15_data https://paddleocr.bj.bcebos.com/dataset/rec_gt_test.txt
```
PaddleOCR 也提供了数据格式转换脚本，可以将官网 label 转换支持的数据格式。 数据转换工具在 `ppocr/utils/gen_label.py`, 这里以训练集为例：
PaddleOCR 也提供了数据格式转换脚本，可以将ICDAR官网 label 转换为PaddleOCR支持的数据格式。 数据转换工具在 `ppocr/utils/gen_label.py`, 这里以训练集为例：
```
# 将官网下载的标签文件转换为 rec_gt_label.txt
python gen_label.py --mode="rec" --input_path="{path/of/origin/label}" --output_label="rec_gt_label.txt"
```
数据样式格式如下，(a)为原始图片,(b)为每张图片对应的 Ground Truth 文本文件：
![](../datasets/icdar_rec.png)
- 多语言数据集
多语言模型的训练数据集均为100w的合成数据，使用了开源合成工具 [text_renderer](https://github.com/Sanster/text_renderer) ，少量的字体可以通过下面两种方式下载。
* [百度网盘](https://pan.baidu.com/s/1bS_u207Rm7YbY33wOECKDA) 提取码：frgi
* [google drive](https://drive.google.com/file/d/18cSWX7wXSy4G0tbKJ0d9PuIaiwRLHpjA/view)
<a name="字典"></a>
1.3 字典
### 1.3 字典
最后需要提供一个字典（{word_dict_name}.txt），使模型在训练时，可以将所有出现的字符映射为字典的索引。
......@@ -152,13 +159,27 @@ PaddleOCR内置了一部分字典，可以按需使用。
并将 `character_type` 设置为 `ch`。
<a name="支持空格"></a>
1.4 添加空格类别
### 1.4 添加空格类别
如果希望支持识别"空格"类别, 请将yml文件中的 `use_space_char` 字段设置为 `True`。
<a name="启动训练"></a>
### 2. 启动训练
## 2. 启动训练
<a name="数据增强"></a>
### 2.1 数据增强
PaddleOCR提供了多种数据增强方式，默认配置文件中已经添加了数据增广。
默认的扰动方式有：颜色空间转换(cvtColor)、模糊(blur)、抖动(jitter)、噪声(Gasuss noise)、随机切割(random crop)、透视(perspective)、颜色反转(reverse)、TIA数据增广。
训练过程中每种扰动方式以40%的概率被选择，具体代码实现请参考：[rec_img_aug.py](../../ppocr/data/imaug/rec_img_aug.py)
*由于OpenCV的兼容性问题，扰动操作暂时只支持Linux*
<a name="通用模型训练"></a>
### 2.2 通用模型训练
PaddleOCR提供了训练脚本、评估脚本和预测脚本，本节将以 CRNN 识别模型为例：
......@@ -178,23 +199,16 @@ tar -xf rec_mv3_none_bilstm_ctc_v2.0_train.tar && rm -rf rec_mv3_none_bilstm_ctc
*如果您安装的是cpu版本，请将配置文件中的 `use_gpu` 字段修改为false*
```
# GPU训练 支持单卡，多卡训练，通过--gpus参数指定卡号
# GPU训练 支持单卡，多卡训练
# 训练icdar15英文数据 训练日志会自动保存为 "{save_model_dir}" 下的train.log
python3 -m paddle.distributed.launch --gpus '0,1,2,3' tools/train.py -c configs/rec/rec_icdar15_train.yml
```
<a name="数据增强"></a>
#### 2.1 数据增强
PaddleOCR提供了多种数据增强方式，如果您希望在训练时加入扰动，请在配置文件中设置 `distort: true`。
默认的扰动方式有：颜色空间转换(cvtColor)、模糊(blur)、抖动(jitter)、噪声(Gasuss noise)、随机切割(random crop)、透视(perspective)、颜色反转(reverse)。
#单卡训练（训练周期长，不建议）
python3 tools/train.py -c configs/rec/rec_icdar15_train.yml
训练过程中每种扰动方式以50%的概率被选择，具体代码实现请参考：[img_tools.py](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/ppocr/data/rec/img_tools.py)
*由于OpenCV的兼容性问题，扰动操作暂时只支持Linux*
#多卡训练，通过--gpus参数指定卡号
python3 -m paddle.distributed.launch --gpus '0,1,2,3' tools/train.py -c configs/rec/rec_icdar15_train.yml
```
<a name="训练"></a>
#### 2.2 训练
PaddleOCR支持训练和评估交替进行, 可以在 `configs/rec/rec_icdar15_train.yml` 中修改 `eval_batch_step` 设置评估频率，默认每500个iter评估一次。评估过程中默认将最佳acc模型，保存为 `output/rec_CRNN/best_accuracy` 。
......@@ -215,6 +229,7 @@ PaddleOCR支持训练和评估交替进行, 可以在 `configs/rec/rec_icdar15_t
| rec_mv3_tps_bilstm_att.yml | CRNN | Mobilenet_v3 | TPS | BiLSTM | att |
| rec_r34_vd_tps_bilstm_att.yml | CRNN | Resnet34_vd | TPS | BiLSTM | att |
| rec_r50fpn_vd_none_srn.yml | SRN | Resnet50_fpn_vd | None | rnn | srn |
| rec_mtb_nrtr.yml | NRTR | nrtr_mtb | None | transformer encoder | transformer decoder |
训练中文数据，推荐使用[rec_chinese_lite_train_v2.0.yml](../../configs/rec/ch_ppocr_v2.0/rec_chinese_lite_train_v2.0.yml)，如您希望尝试其他算法在中文数据集上的效果，请参考下列说明修改配置文件：
......@@ -282,85 +297,12 @@ Eval:
```
**注意，预测/评估时的配置文件请务必与训练一致。**
<a name="小语种"></a>
#### 2.3 小语种
<a name="多语言模型训练"></a>
### 2.3 多语言模型训练
PaddleOCR目前已支持80种（除中文外）语种识别，`configs/rec/multi_languages` 路径下提供了一个多语言的配置文件模版: [rec_multi_language_lite_train.yml](../../configs/rec/multi_language/rec_multi_language_lite_train.yml)。
您有两种方式创建所需的配置文件：
1. 通过脚本自动生成
[generate_multi_language_configs.py](../../configs/rec/multi_language/generate_multi_language_configs.py) 可以帮助您生成多语言模型的配置文件
- 以意大利语为例，如果您的数据是按如下格式准备的：
```
|-train_data
|- it_train.txt # 训练集标签
|- it_val.txt # 验证集标签
|- data
|- word_001.jpg
|- word_002.jpg
|- word_003.jpg
| ...
```
可以使用默认参数，生成配置文件：
```bash
# 该代码需要在指定目录运行
cd PaddleOCR/configs/rec/multi_language/
# 通过-l或者--language参数设置需要生成的语种的配置文件，该命令会将默认参数写入配置文件
python3 generate_multi_language_configs.py -l it
```
- 如果您的数据放置在其他位置，或希望使用自己的字典，可以通过指定相关参数来生成配置文件:
```bash
# -l或者--language字段是必须的
# --train修改训练集，--val修改验证集，--data_dir修改数据集目录，--dict修改字典路径， -o修改对应默认参数
cd PaddleOCR/configs/rec/multi_language/
python3 generate_multi_language_configs.py -l it \ # 语种
--train {path/of/train_label.txt} \ # 训练标签文件的路径
--val {path/of/val_label.txt} \ # 验证集标签文件的路径
--data_dir {train_data/path} \ # 训练数据的根目录
--dict {path/of/dict} \ # 字典文件路径
-o Global.use_gpu=False # 是否使用gpu
...
```
意大利文由拉丁字母组成，因此执行完命令后会得到名为 rec_latin_lite_train.yml 的配置文件。
2. 手动修改配置文件
您也可以手动修改模版中的以下几个字段:
```
Global:
use_gpu: True
epoch_num: 500
...
character_type: it # 需要识别的语种
character_dict_path: {path/of/dict} # 字典文件所在路径
Train:
dataset:
name: SimpleDataSet
data_dir: train_data/ # 数据存放根目录
label_file_list: ["./train_data/train_list.txt"] # 训练集label路径
...
Eval:
dataset:
name: SimpleDataSet
data_dir: train_data/ # 数据存放根目录
label_file_list: ["./train_data/val_list.txt"] # 验证集label路径
...
```
目前PaddleOCR支持的多语言算法有：
按语系划分，目前PaddleOCR支持的语种有：
| 配置文件 | 算法名称 | backbone | trans | seq | pred | language | character_type |
| :--------: | :-------: | :-------: | :-------: | :-----: | :-----: | :-----: | :-----: |
......@@ -377,10 +319,6 @@ PaddleOCR目前已支持80种（除中文外）语种识别，`configs/rec/multi
更多支持语种请参考: [多语言模型](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.1/doc/doc_ch/multi_languages.md#%E8%AF%AD%E7%A7%8D%E7%BC%A9%E5%86%99)
多语言模型训练方式与中文模型一致，训练数据集均为100w的合成数据，少量的字体可以通过下面两种方式下载。
* [百度网盘](https://pan.baidu.com/s/1bS_u207Rm7YbY33wOECKDA)。提取码：frgi。
* [google drive](https://drive.google.com/file/d/18cSWX7wXSy4G0tbKJ0d9PuIaiwRLHpjA/view)
如您希望在现有模型效果的基础上调优，请参考下列说明修改配置文件：
以 `rec_french_lite_train` 为例：
......@@ -416,7 +354,7 @@ Eval:
...
```
<a name="评估"></a>
### 3 评估
## 3 评估
评估数据集可以通过 `configs/rec/rec_icdar15_train.yml` 修改Eval中的 `label_file_path` 设置。
......@@ -426,14 +364,29 @@ python3 -m paddle.distributed.launch --gpus '0' tools/eval.py -c configs/rec/rec
```
<a name="预测"></a>
### 4 预测
<a name="训练引擎预测"></a>
#### 4.1 训练引擎的预测
## 4 预测
使用 PaddleOCR 训练好的模型，可以通过以下脚本进行快速预测。
默认预测图片存储在 `infer_img` 里，通过 `-o Global.checkpoints` 指定权重：
默认预测图片存储在 `infer_img` 里，通过 `-o Global.checkpoints` 加载训练好的参数文件：
根据配置文件中设置的的 `save_model_dir` 和 `save_epoch_step` 字段，会有以下几种参数被保存下来：
```
output/rec/
├── best_accuracy.pdopt
├── best_accuracy.pdparams
├── best_accuracy.states
├── config.yml
├── iter_epoch_3.pdopt
├── iter_epoch_3.pdparams
├── iter_epoch_3.states
├── latest.pdopt
├── latest.pdparams
├── latest.states
└── train.log
```
其中 best_accuracy.* 是评估集上的最优模型；iter_epoch_x.* 是以 `save_epoch_step` 为间隔保存下来的模型；latest.* 是最后一个epoch的模型。
```
# 预测英文结果
......@@ -469,3 +422,39 @@ python3 tools/infer_rec.py -c configs/rec/ch_ppocr_v2.0/rec_chinese_lite_train_v
infer_img: doc/imgs_words/ch/word_1.jpg
result: ('韩国小馆', 0.997218)
```
<a name="Inference"></a>
## 5. 转Inference模型测试
识别模型转inference模型与检测的方式相同，如下：
```
# -c 后面设置训练算法的yml配置文件
# -o 配置可选参数
# Global.pretrained_model 参数设置待转换的训练模型地址，不用添加文件后缀 .pdmodel，.pdopt或.pdparams。
# Global.save_inference_dir参数设置转换的模型将保存的地址。
python3 tools/export_model.py -c configs/rec/ch_ppocr_v2.0/rec_chinese_lite_train_v2.0.yml -o Global.pretrained_model=./ch_lite/ch_ppocr_mobile_v2.0_rec_train/best_accuracy Global.save_inference_dir=./inference/rec_crnn/
```
**注意：**如果您是在自己的数据集上训练的模型，并且调整了中文字符的字典文件，请注意修改配置文件中的`character_dict_path`是否是所需要的字典文件。
转换成功后，在目录下有三个文件：
```
/inference/rec_crnn/
├── inference.pdiparams # 识别inference模型的参数文件
├── inference.pdiparams.info # 识别inference模型的参数信息，可忽略
└── inference.pdmodel # 识别inference模型的program文件
```
- 自定义模型推理
如果训练时修改了文本的字典，在使用inference模型预测时，需要通过`--rec_char_dict_path`指定使用的字典路径，并且设置 `rec_char_type=ch`
```
python3 tools/infer/predict_rec.py --image_dir="./doc/imgs_words_en/word_336.png" --rec_model_dir="./your inference model" --rec_image_shape="3, 32, 100" --rec_char_type="ch" --rec_char_dict_path="your text dict path"
```
doc/doc_ch/training.md 0 → 100644
浏览文件 @ a8a9b2e5
# 模型训练
本文将介绍模型训练时需掌握的基本概念，和训练时的调优方法。
同时会简单介绍PaddleOCR模型训练数据的组成部分，以及如何在垂类场景中准备数据finetune模型。
- [1. 基本概念](#基本概念)
* [1.1 学习率](#学习率)
* [1.2 正则化](#正则化)
* [1.3 评估指标](#评估指标)
- [2. 数据与垂类场景](#数据与垂类场景)
* [2.1 训练数据](#训练数据)
* [2.2 垂类场景](#垂类场景)
* [2.3 自己构建数据集](#自己构建数据集)
* [3. 常见问题](#常见问题)
<a name="基本概念"></a>
## 1. 基本概念
OCR(Optical Character Recognition,光学字符识别)是指对图像进行分析识别处理，获取文字和版面信息的过程，是典型的计算机视觉任务，
通常由文本检测和文本识别两个子任务构成。
模型调优时需要关注以下参数：
<a name="学习率"></a>
### 1.1 学习率
学习率是训练神经网络的重要超参数之一，它代表在每一次迭代中梯度向损失函数最优解移动的步长。
在PaddleOCR中提供了多种学习率更新策略,可以通过配置文件修改，例如：
```
Optimizer:
...
lr:
name: Piecewise
decay_epochs : [700, 800]
values : [0.001, 0.0001]
warmup_epoch: 5
```
Piecewise 代表分段常数衰减，在不同的学习阶段指定不同的学习率，在每段内学习率相同。
warmup_epoch 代表在前5个epoch中，学习率将逐渐从0增加到base_lr。全部策略可以参考代码[learning_rate.py](../../ppocr/optimizer/learning_rate.py) 。
<a name="正则化"></a>
### 1.2 正则化
正则化可以有效的避免算法过拟合，PaddleOCR中提供了L1、L2正则方法，L1 和 L2 正则化是最常用的正则化方法。L1 正则化向目标函数添加正则化项，以减少参数的绝对值总和；而 L2 正则化中，添加正则化项的目的在于减少参数平方的总和。配置方法如下：
```
Optimizer:
...
regularizer:
name: L2
factor: 2.0e-05
```
<a name="评估指标"></a>
### 1.3 评估指标
（1）检测阶段：先按照检测框和标注框的IOU评估，IOU大于某个阈值判断为检测准确。这里检测框和标注框不同于一般的通用目标检测框，是采用多边形进行表示。检测准确率：正确的检测框个数在全部检测框的占比，主要是判断检测指标。检测召回率：正确的检测框个数在全部标注框的占比，主要是判断漏检的指标。
（2）识别阶段： 字符识别准确率，即正确识别的文本行占标注的文本行数量的比例，只有整行文本识别对才算正确识别。
（3）端到端统计： 端对端召回率：准确检测并正确识别文本行在全部标注文本行的占比； 端到端准确率：准确检测并正确识别文本行在 检测到的文本行数量 的占比； 准确检测的标准是检测框与标注框的IOU大于某个阈值，正确识别的的检测框中的文本与标注的文本相同。
<a name="数据与垂类场景"></a>
## 2. 数据与垂类场景
<a name="训练数据"></a>
### 2.1 训练数据
目前开源的模型，数据集和量级如下：
- 检测：
- 英文数据集，ICDAR2015
- 中文数据集，LSVT街景数据集训练数据3w张图片
- 识别：
- 英文数据集，MJSynth和SynthText合成数据，数据量上千万。
- 中文数据集，LSVT街景数据集根据真值将图crop出来，并进行位置校准，总共30w张图像。此外基于LSVT的语料，合成数据500w。
- 小语种数据集，使用不同语料和字体，分别生成了100w合成数据集，并使用ICDAR-MLT作为验证集。
其中，公开数据集都是开源的，用户可自行搜索下载，也可参考[中文数据集](./datasets.md)，合成数据暂不开源，用户可使用开源合成工具自行合成，可参考的合成工具包括[text_renderer](https://github.com/Sanster/text_renderer) 、[SynthText](https://github.com/ankush-me/SynthText) 、[TextRecognitionDataGenerator](https://github.com/Belval/TextRecognitionDataGenerator) 等。
<a name="垂类场景"></a>
### 2.2 垂类场景
PaddleOCR主要聚焦通用OCR，如果有垂类需求，您可以用PaddleOCR+垂类数据自己训练；
如果缺少带标注的数据，或者不想投入研发成本，建议直接调用开放的API，开放的API覆盖了目前比较常见的一些垂类。
<a name="自己构建数据集"></a>
### 2.3 自己构建数据集
在构建数据集时有几个经验可供参考：
（1） 训练集的数据量：
a. 检测需要的数据相对较少，在PaddleOCR模型的基础上进行Fine-tune，一般需要500张可达到不错的效果。
b. 识别分英文和中文，一般英文场景需要几十万数据可达到不错的效果，中文则需要几百万甚至更多。
（2）当训练数据量少时，可以尝试以下三种方式获取更多的数据：
a. 人工采集更多的训练数据，最直接也是最有效的方式。
b. 基于PIL和opencv基本图像处理或者变换。例如PIL中ImageFont, Image, ImageDraw三个模块将文字写到背景中，opencv的旋转仿射变换，高斯滤波等。
c. 利用数据生成算法合成数据，例如pix2pix或StyleText等算法。
<a name="常见问题"></a>
## 3. 常见问题
**Q**：训练CRNN识别时，如何选择合适的网络输入shape？
A：一般高度采用32，最长宽度的选择，有两种方法：
（1）统计训练样本图像的宽高比分布。最大宽高比的选取考虑满足80%的训练样本。
（2）统计训练样本文字数目。最长字符数目的选取考虑满足80%的训练样本。然后中文字符长宽比近似认为是1，英文认为3：1，预估一个最长宽度。
**Q**：识别训练时，训练集精度已经到达90了，但验证集精度一直在70，涨不上去怎么办？
A：训练集精度90，测试集70多的话，应该是过拟合了，有两个可尝试的方法：
（1）加入更多的增广方式或者调大增广prob的[概率](https://github.com/PaddlePaddle/PaddleOCR/blob/dygraph/ppocr/data/imaug/rec_img_aug.py#L341)，默认为0.4。
（2）调大系统的[l2 dcay值](https://github.com/PaddlePaddle/PaddleOCR/blob/a501603d54ff5513fc4fc760319472e59da25424/configs/rec/ch_ppocr_v1.1/rec_chinese_lite_train_v1.1.yml#L47)
**Q**: 识别模型训练时，loss能正常下降，但acc一直为0
A：识别模型训练初期acc为0是正常的，多训一段时间指标就上来了。
doc/doc_ch/update.md
浏览文件 @ a8a9b2e5
# 更新
- 2021.9.7 发布PaddleOCR v2.3，发布[PP-OCRv2](#PP-OCRv2)，CPU推理速度相比于PP-OCR server提升220%；效果相比于PP-OCR mobile 提升7%。
- 2021.8.3 发布PaddleOCR v2.2，新增文档结构分析[PP-Structure](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.2/ppstructure/README_ch.md)工具包，支持版面分析与表格识别（含Excel导出）。
- 2021.6.29 [FAQ](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.2/doc/doc_ch/FAQ.md)新增5个高频问题，总数248个，每周一都会更新，欢迎大家持续关注。
- 2021.4.8 release 2.1版本，新增AAAI 2021论文[端到端识别算法PGNet](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.2/doc/doc_ch/pgnet.md)开源，[多语言模型](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.2/doc/doc_ch/multi_languages.md)支持种类增加到80+。
- 2020.12.15 更新数据合成工具[Style-Text](../../StyleText/README_ch.md)，可以批量合成大量与目标场景类似的图像，在多个场景验证，效果明显提升。
- 2020.12.07 [FAQ](../../doc/doc_ch/FAQ.md)新增5个高频问题，总数124个，并且计划以后每周一都会更新，欢迎大家持续关注。
- 2020.11.25 更新半自动标注工具[PPOCRLabel](../../PPOCRLabel/README_ch.md)，辅助开发者高效完成标注任务，输出格式与PP-OCR训练任务完美衔接。
......
doc/doc_ch/visualization.md
浏览文件 @ a8a9b2e5
# 效果展示
<a name="超轻量PP-OCRv2效果展示"></a>
## 超轻量PP-OCRv2效果展示
<img src="../imgs_results/PP-OCRv2/PP-OCRv2-pic001.jpg" width="800">
<img src="../imgs_results/PP-OCRv2/PP-OCRv2-pic002.jpg" width="800">
<img src="../imgs_results/PP-OCRv2/PP-OCRv2-pic003.jpg" width="800">
<a name="超轻量ppocr_server_2.0效果展示"></a>
## 通用ppocr_server_2.0 效果展示
## 通用PP-OCR server 效果展示
<div align="center">
<img src="../imgs_results/ch_ppocr_mobile_v2.0/00006737.jpg" width="800">
......@@ -10,8 +16,6 @@
<img src="../imgs_results/ch_ppocr_mobile_v2.0/00018069.jpg" width="800">
<img src="../imgs_results/ch_ppocr_mobile_v2.0/00056221.jpg" width="800">
<img src="../imgs_results/ch_ppocr_mobile_v2.0/00057937.jpg" width="800">
<img src="../imgs_results/ch_ppocr_mobile_v2.0/00059985.jpg" width="800">
<img src="../imgs_results/ch_ppocr_mobile_v2.0/00111002.jpg" width="800">
<img src="../imgs_results/ch_ppocr_mobile_v2.0/00077949.jpg" width="800">
<img src="../imgs_results/ch_ppocr_mobile_v2.0/00207393.jpg" width="800">
</div>
......
doc/doc_ch/whl.md
浏览文件 @ a8a9b2e5
......@@ -210,7 +210,7 @@ paddleocr --image_dir PaddleOCR/doc/imgs/11.jpg --use_angle_cls true
```bash
[[[24.0, 36.0], [304.0, 34.0], [304.0, 72.0], [24.0, 74.0]], ['纯臻营养护发素', 0.964739]]
[[[24.0, 80.0], [172.0, 80.0], [172.0, 104.0], [24.0, 104.0]], ['产品信息/参数', 0.98069626]]
[[[24.0, 109.0], [333.0, 109.0], [333.0, 136.0], [24.0, 136.0]], ['（45元/每公斤，100公斤起订）', 0.9676722]]
[[[24.0, 109.0], [333.0, 109.0], [333.0, 136.0], [24.0, 136.0]], ['（45元/每公斤，100公斤起订）', 0.9676722]]µ
......
```
......
doc/doc_en/algorithm_overview_en.md
浏览文件 @ a8a9b2e5
......@@ -46,6 +46,7 @@ PaddleOCR open-source text recognition algorithms list:
- [x] STAR-Net([paper](http://www.bmva.org/bmvc/2016/papers/paper043/index.html))[11]
- [x] RARE([paper](https://arxiv.org/abs/1603.03915v1))[12]
- [x] SRN([paper](https://arxiv.org/abs/2003.12294))[5]
- [x] NRTR([paper](https://arxiv.org/abs/1806.00926v2))
Refer to [DTRB](https://arxiv.org/abs/1904.01906), the training and evaluation result of these above text recognition (using MJSynth and SynthText for training, evaluate on IIIT, SVT, IC03, IC13, IC15, SVTP, CUTE) is as follow:
......@@ -60,5 +61,6 @@ Refer to [DTRB](https://arxiv.org/abs/1904.01906), the training and evaluation r
|RARE|MobileNetV3|82.5%|rec_mv3_tps_bilstm_att |[Download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_mv3_tps_bilstm_att_v2.0_train.tar)|
|RARE|Resnet34_vd|83.6%|rec_r34_vd_tps_bilstm_att |[Download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_r34_vd_tps_bilstm_att_v2.0_train.tar)|
|SRN|Resnet50_vd_fpn| 88.52% | rec_r50fpn_vd_none_srn |[Download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_r50_vd_srn_train.tar)|
|NRTR|NRTR_MTB| 84.3% | rec_mtb_nrtr | [Download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_mtb_nrtr_train.tar) |
Please refer to the document for training guide and use of PaddleOCR text recognition algorithms [Text recognition model training/evaluation/prediction](./recognition_en.md)
doc/doc_en/config_en.md
浏览文件 @ a8a9b2e5
......@@ -51,7 +51,7 @@ Take rec_chinese_lite_train_v2.0.yml as an example
### Architecture ([ppocr/modeling](../../ppocr/modeling))
In ppocr, the network is divided into four stages: Transform, Backbone, Neck and Head
In PaddleOCR, the network is divided into four stages: Transform, Backbone, Neck and Head
| Parameter | Use | Defaults | Note |
| :---------------------: | :---------------------: | :--------------: | :--------------------: |
......@@ -120,3 +120,108 @@ In ppocr, the network is divided into four stages: Transform, Backbone, Neck and
| batch_size_per_card | Single card batch size during training | 256 | \ |
| drop_last | Whether to discard the last incomplete mini-batch because the number of samples in the data set cannot be divisible by batch_size | True | \ |
| num_workers | The number of sub-processes used to load data, if it is 0, the sub-process is not started, and the data is loaded in the main process | 8 | \ |
## 3. MULTILINGUAL CONFIG FILE GENERATION
PaddleOCR currently supports 80 (except Chinese) language recognition. A multi-language configuration file template is
provided under the path `configs/rec/multi_languages`: [rec_multi_language_lite_train.yml](../../configs/rec/multi_language/rec_multi_language_lite_train.yml)。
There are two ways to create the required configuration file:：
1. Automatically generated by script
[generate_multi_language_configs.py](../../configs/rec/multi_language/generate_multi_language_configs.py) Can help you generate configuration files for multi-language models
- Take Italian as an example, if your data is prepared in the following format:
```
|-train_data
|- it_train.txt # train_set label
|- it_val.txt # val_set label
|- data
|- word_001.jpg
|- word_002.jpg
|- word_003.jpg
| ...
```
You can use the default parameters to generate a configuration file:
```bash
# The code needs to be run in the specified directory
cd PaddleOCR/configs/rec/multi_language/
# Set the configuration file of the language to be generated through the -l or --language parameter.
# This command will write the default parameters into the configuration file
python3 generate_multi_language_configs.py -l it
```
- If your data is placed in another location, or you want to use your own dictionary, you can generate the configuration file by specifying the relevant parameters:
```bash
# -l or --language field is required
# --train to modify the training set
# --val to modify the validation set
# --data_dir to modify the data set directory
# --dict to modify the dict path
# -o to modify the corresponding default parameters
cd PaddleOCR/configs/rec/multi_language/
python3 generate_multi_language_configs.py -l it \ # language
--train {path/of/train_label.txt} \ # path of train_label
--val {path/of/val_label.txt} \ # path of val_label
--data_dir {train_data/path} \ # root directory of training data
--dict {path/of/dict} \ # path of dict
-o Global.use_gpu=False # whether to use gpu
...
```
Italian is made up of Latin letters, so after executing the command, you will get the rec_latin_lite_train.yml.
2. Manually modify the configuration file
You can also manually modify the following fields in the template:
```
Global:
use_gpu: True
epoch_num: 500
...
character_type: it # language
character_dict_path: {path/of/dict} # path of dict
Train:
dataset:
name: SimpleDataSet
data_dir: train_data/ # root directory of training data
label_file_list: ["./train_data/train_list.txt"] # train label path
...
Eval:
dataset:
name: SimpleDataSet
data_dir: train_data/ # root directory of val data
label_file_list: ["./train_data/val_list.txt"] # val label path
...
```
Currently, the multi-language algorithms supported by PaddleOCR are:
| Configuration file | Algorithm name | backbone | trans | seq | pred | language | character_type |
| :--------: | :-------: | :-------: | :-------: | :-----: | :-----: | :-----: | :-----: |
| rec_chinese_cht_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | chinese traditional | chinese_cht|
| rec_en_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | English(Case sensitive) | EN |
| rec_french_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | French | french |
| rec_ger_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | German | german |
| rec_japan_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | Japanese | japan |
| rec_korean_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | Korean | korean |
| rec_latin_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | Latin | latin |
| rec_arabic_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | arabic | ar |
| rec_cyrillic_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | cyrillic | cyrillic |
| rec_devanagari_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | devanagari | devanagari |
For more supported languages, please refer to : [Multi-language model](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.1/doc/doc_en/multi_languages_en.md#4-support-languages-and-abbreviations)
The multi-language model training method is the same as the Chinese model. The training data set is 100w synthetic data. A small amount of fonts and test data can be downloaded using the following two methods.
* [Baidu Netdisk](https://pan.baidu.com/s/1bS_u207Rm7YbY33wOECKDA),Extraction code:frgi.
* [Google drive](https://drive.google.com/file/d/18cSWX7wXSy4G0tbKJ0d9PuIaiwRLHpjA/view)
doc/doc_en/detection_en.md
浏览文件 @ a8a9b2e5
# TEXT DETECTION
# CONTENT
- [Paste Your Document In Here](#paste-your-document-in-here)
- [1. TEXT DETECTION](#1-text-detection)
* [1.1 DATA PREPARATION](#11-data-preparation)
* [1.2 DOWNLOAD PRETRAINED MODEL](#12-download-pretrained-model)
* [1.3 START TRAINING](#13-start-training)
* [1.4 LOAD TRAINED MODEL AND CONTINUE TRAINING](#14-load-trained-model-and-continue-training)
* [1.5 TRAINING WITH NEW BACKBONE](#15-training-with-new-backbone)
* [1.6 EVALUATION](#16-evaluation)
* [1.7 TEST](#17-test)
* [1.8 INFERENCE MODEL PREDICTION](#18-inference-model-prediction)
- [2. FAQ](#2-faq)
# 1. TEXT DETECTION
This section uses the icdar2015 dataset as an example to introduce the training, evaluation, and testing of the detection model in PaddleOCR.
## DATA PREPARATION
The icdar2015 dataset can be obtained from [official website](https://rrc.cvc.uab.es/?ch=4&com=downloads). Registration is required for downloading.
## 1.1 DATA PREPARATION
The icdar2015 dataset contains train set which has 1000 images obtained with wearable cameras and test set which has 500 images obtained with wearable cameras. The icdar2015 can be obtained from [official website](https://rrc.cvc.uab.es/?ch=4&com=downloads). Registration is required for downloading.
After registering and logging in, download the part marked in the red box in the figure below. And, the content downloaded by `Training Set Images` should be saved as the folder `icdar_c4_train_imgs`, and the content downloaded by `Test Set Images` is saved as the folder `ch4_test_images`
<p align="center">
<img src="../datasets/ic15_location_download.png" align="middle" width = "700"/>
<p align="center">
Decompress the downloaded dataset to the working directory, assuming it is decompressed under PaddleOCR/train_data/. In addition, PaddleOCR organizes many scattered annotation files into two separate annotation files for train and test respectively, which can be downloaded by wget:
```shell
......@@ -36,10 +59,11 @@ The `points` in the dictionary represent the coordinates (x, y) of the four poin
If you want to train PaddleOCR on other datasets, please build the annotation file according to the above format.
## TRAINING
## 1.2 DOWNLOAD PRETRAINED MODEL
First download the pretrained model. The detection model of PaddleOCR currently supports 3 backbones, namely MobileNetV3, ResNet18_vd and ResNet50_vd. You can use the model in [PaddleClas](https://github.com/PaddlePaddle/PaddleClas/tree/release/2.0/ppcls/modeling/architectures) to replace backbone according to your needs.
And the responding download link of backbone pretrain weights can be found in (https://github.com/PaddlePaddle/PaddleClas/blob/release%2F2.0/README_cn.md#resnet%E5%8F%8A%E5%85%B6vd%E7%B3%BB%E5%88%97).
First download the pretrained model. The detection model of PaddleOCR currently supports 3 backbones, namely MobileNetV3, ResNet18_vd and ResNet50_vd. You can use the model in [PaddleClas](https://github.com/PaddlePaddle/PaddleClas/tree/develop/ppcls/modeling/architectures) to replace backbone according to your needs.
And the responding download link of backbone pretrain weights can be found in [PaddleClas repo](https://github.com/PaddlePaddle/PaddleClas#mobile-series).
```shell
cd PaddleOCR/
# Download the pre-trained model of MobileNetV3
......@@ -49,11 +73,13 @@ wget -P ./pretrain_models/ https://paddle-imagenet-models-name.bj.bcebos.com/dyg
# or, download the pre-trained model of ResNet50_vd
wget -P ./pretrain_models/ https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet50_vd_ssld_pretrained.pdparams
```
#### START TRAINING
## 1.3 START TRAINING
*If CPU version installed, please set the parameter `use_gpu` to `false` in the configuration.*
```shell
python3 tools/train.py -c configs/det/det_mv3_db.yml
python3 tools/train.py -c configs/det/det_mv3_db.yml \
-o Global.pretrain_weights=./pretrain_models/MobileNetV3_large_x0_5_pretrained
```
In the above instruction, use `-c` to select the training to use the `configs/det/det_db_mv3.yml` configuration file.
......@@ -62,16 +88,17 @@ For a detailed explanation of the configuration file, please refer to [config](.
You can also use `-o` to change the training parameters without modifying the yml file. For example, adjust the training learning rate to 0.0001
```shell
# single GPU training
python3 tools/train.py -c configs/det/det_mv3_db.yml -o Optimizer.base_lr=0.0001
python3 tools/train.py -c configs/det/det_mv3_db.yml -o \
Global.pretrain_weights=./pretrain_models/MobileNetV3_large_x0_5_pretrained \
Optimizer.base_lr=0.0001
# multi-GPU training
# Set the GPU ID used by the '--gpus' parameter.
python3 -m paddle.distributed.launch --gpus '0,1,2,3' tools/train.py -c configs/det/det_mv3_db.yml -o Optimizer.base_lr=0.0001
python3 -m paddle.distributed.launch --gpus '0,1,2,3' tools/train.py -c configs/det/det_mv3_db.yml -o Global.pretrain_weights=./pretrain_models/MobileNetV3_large_x0_5_pretrained
```
#### load trained model and continue training
## 1.4 LOAD TRAINED MODEL AND CONTINUE TRAINING
If you expect to load trained model and continue the training again, you can specify the parameter `Global.checkpoints` as the model path to be loaded.
For example:
......@@ -82,9 +109,59 @@ python3 tools/train.py -c configs/det/det_mv3_db.yml -o Global.checkpoints=./you
**Note**: The priority of `Global.checkpoints` is higher than that of `Global.pretrain_weights`, that is, when two parameters are specified at the same time, the model specified by `Global.checkpoints` will be loaded first. If the model path specified by `Global.checkpoints` is wrong, the one specified by `Global.pretrain_weights` will be loaded.
## EVALUATION
## 1.5 TRAINING WITH NEW BACKBONE
The network part completes the construction of the network, and PaddleOCR divides the network into four parts, which are under [ppocr/modeling](../../ppocr/modeling). The data entering the network will pass through these four parts in sequence(transforms->backbones->
necks->heads).
```bash
├── architectures # Code for building network
├── transforms # Image Transformation Module
├── backbones # Feature extraction module
├── necks # Feature enhancement module
└── heads # Output module
```
If the Backbone to be replaced has a corresponding implementation in PaddleOCR, you can directly modify the parameters in the `Backbone` part of the configuration yml file.
However, if you want to use a new Backbone, an example of replacing the backbones is as follows:
1. Create a new file under the [ppocr/modeling/backbones](../../ppocr/modeling/backbones) folder, such as my_backbone.py.
2. Add code in the my_backbone.py file, the sample code is as follows:
```python
import paddle
import paddle.nn as nn
import paddle.nn.functional as F
class MyBackbone(nn.Layer):
def __init__(self, *args, **kwargs):
super(MyBackbone, self).__init__()
# your init code
self.conv = nn.xxxx
def forward(self, inputs):
# your network forward
y = self.conv(inputs)
return y
```
3. Import the added module in the [ppocr/modeling/backbones/\__init\__.py](../../ppocr/modeling/backbones/__init__.py) file.
PaddleOCR calculates three indicators for evaluating performance of OCR detection task: Precision, Recall, and Hmean.
After adding the four-part modules of the network, you only need to configure them in the configuration file to use, such as:
```yaml
Backbone:
name: MyBackbone
args1: args1
```
**NOTE**: More details about replace Backbone and other mudule can be found in [doc](add_new_algorithm_en.md).
## 1.6 EVALUATION
PaddleOCR calculates three indicators for evaluating performance of OCR detection task: Precision, Recall, and Hmean(F-Score).
Run the following code to calculate the evaluation indicators. The result will be saved in the test result file specified by `save_res_path` in the configuration file `det_db_mv3.yml`
......@@ -95,10 +172,9 @@ The model parameters during training are saved in the `Global.save_model_dir` di
python3 tools/eval.py -c configs/det/det_mv3_db.yml -o Global.checkpoints="{path/to/weights}/best_accuracy" PostProcess.box_thresh=0.6 PostProcess.unclip_ratio=1.5
```
* Note: `box_thresh` and `unclip_ratio` are parameters required for DB post-processing, and not need to be set when evaluating the EAST and SAST model.
* Note: `box_thresh` and `unclip_ratio` are parameters required for DB post-processing, and not need to be set when evaluating the EAST model.
## TEST
## 1.7 TEST
Test the detection result on a single image:
```shell
......@@ -107,7 +183,7 @@ python3 tools/infer_det.py -c configs/det/det_mv3_db.yml -o Global.infer_img="./
When testing the DB model, adjust the post-processing threshold:
```shell
python3 tools/infer_det.py -c configs/det/det_mv3_db.yml -o Global.infer_img="./doc/imgs_en/img_10.jpg" Global.pretrained_model="./output/det_db/best_accuracy" PostProcess.box_thresh=0.6 PostProcess.unclip_ratio=1.5
python3 tools/infer_det.py -c configs/det/det_mv3_db.yml -o Global.infer_img="./doc/imgs_en/img_10.jpg" Global.pretrained_model="./output/det_db/best_accuracy" PostProcess.box_thresh=0.6 PostProcess.unclip_ratio=2.0
```
......@@ -115,3 +191,33 @@ Test the detection result on all images in the folder:
```shell
python3 tools/infer_det.py -c configs/det/det_mv3_db.yml -o Global.infer_img="./doc/imgs_en/" Global.pretrained_model="./output/det_db/best_accuracy"
```
## 1.8 INFERENCE MODEL PREDICTION
The inference model (the model saved by `paddle.jit.save`) is generally a solidified model saved after the model training is completed, and is mostly used to give prediction in deployment.
The model saved during the training process is the checkpoints model, which saves the parameters of the model and is mostly used to resume training.
Compared with the checkpoints model, the inference model will additionally save the structural information of the model. Therefore, it is easier to deploy because the model structure and model parameters are already solidified in the inference model file, and is suitable for integration with actual systems.
Firstly, we can convert DB trained model to inference model:
```shell
python3 tools/export_model.py -c configs/det/det_mv3_db.yml -o Global.pretrained_model="./output/det_db/best_accuracy" Global.save_inference_dir="./output/det_db_inference/"
```
The detection inference model prediction：
```shell
python3 tools/infer/predict_det.py --det_algorithm="DB" --det_model_dir="./output/det_db_inference/" --image_dir="./doc/imgs/" --use_gpu=True
```
If it is other detection algorithms, such as the EAST, the det_algorithm parameter needs to be modified to EAST, and the default is the DB algorithm:
```shell
python3 tools/infer/predict_det.py --det_algorithm="EAST" --det_model_dir="./output/det_db_inference/" --image_dir="./doc/imgs/" --use_gpu=True
```
# 2. FAQ
Q1: The prediction results of trained model and inference model are inconsistent?
**A**: Most of the problems are caused by the inconsistency of the pre-processing and post-processing parameters during the prediction of the trained model and the pre-processing and post-processing parameters during the prediction of the inference model. Taking the model trained by the det_mv3_db.yml configuration file as an example, the solution to the problem of inconsistent prediction results between the training model and the inference model is as follows:
- Check whether the [trained model preprocessing](https://github.com/PaddlePaddle/PaddleOCR/blob/c1ed243fb68d5d466258243092e56cbae32e2c14/configs/det/det_mv3_db.yml#L116) is consistent with the prediction [preprocessing function of the inference model](https://github.com/PaddlePaddle/PaddleOCR/blob/c1ed243fb68d5d466258243092e56cbae32e2c14/tools/infer/predict_det.py#L42). When the algorithm is evaluated, the input image size will affect the accuracy. In order to be consistent with the paper, the image is resized to [736, 1280] in the training icdar15 configuration file, but there is only a set of default parameters when the inference model predicts, which will be considered To predict the speed problem, the longest side of the image is limited to 960 for resize by default. The preprocessing function of the training model preprocessing and the inference model is located in [ppocr/data/imaug/operators.py](https://github.com/PaddlePaddle/PaddleOCR/blob/c1ed243fb68d5d466258243092e56cbae32e2c14/ppocr/data/imaug/operators.py#L147)
- Check whether the [post-processing of the trained model](https://github.com/PaddlePaddle/PaddleOCR/blob/c1ed243fb68d5d466258243092e56cbae32e2c14/configs/det/det_mv3_db.yml#L51) is consistent with the [post-processing parameters of the inference](https://github.com/PaddlePaddle/PaddleOCR/blob/c1ed243fb68d5d466258243092e56cbae32e2c14/tools/infer/utility.py#L50).
doc/doc_en/environment_en.md 0 → 100644
浏览文件 @ a8a9b2e5
# Environment Preparation
* [1. Python Environment Setup](#1)
+ [1.1 Windows](#1.1)
+ [1.2 Mac](#1.2)
+ [1.3 Linux](#1.3)
* [2. Install PaddlePaddle 2.0](#2)
<a name="1"></a>
## 1. Python Environment Setup
<a name="1.1"></a>
### 1.1 Windows
#### 1.1.1 Install Anaconda
- Note: To use paddlepaddle you need to install python environment first, here we choose python integrated environment Anaconda toolkit
- Anaconda is a common python package manager
- After installing Anaconda, you can install the python environment, as well as numpy and other required toolkit environment.
- Anaconda download.
- Address: https://mirrors.tuna.tsinghua.edu.cn/anaconda/archive/?C=M&O=D
- Most Win10 computers are 64-bit operating systems, choose x86_64 version; if the computer is a 32-bit operating system, choose x86.exe
<img src="../install/windows/Anaconda_download.png" alt="anaconda download" width="800" align="center "/>
- After the download is complete, double-click the installer to enter the graphical interface
- The default installation location is C drive, it is recommended to change the installation location to D drive.
<img src="../install/windows/anaconda_install_folder.png" alt="install config" width="500" align=" left"/>
- Check conda to add environment variables and ignore the warning that
<img src="../install/windows/anaconda_install_env.png" alt="add conda to path" width="500" align="center"/>
#### 1.1.2 Opening the terminal and creating the conda environment
- Open Anaconda Prompt terminal: bottom left Windows Start Menu -> Anaconda3 -> Anaconda Prompt start console
<img src="../install/windows/anaconda_prompt.png" alt="anaconda download" width="300" align="center"/>
- Create a new conda environment
```shell
# Enter the following command at the command line to create an environment named paddle_env
# Here to speed up the download, use the Tsinghua source
conda create --name paddle_env python=3.8 --channel https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/ # This is a one line command
```
This command will create an executable environment named paddle_env with python version 3.8, which will take a while depending on the network status
The command line will then output a prompt, type y and enter to continue the installation
<img src="../install/windows/conda_new_env.png" alt="conda create" width="700" align="center"/>
- To activate the conda environment you just created, enter the following command at the command line.
```shell
# Activate the paddle_env environment
conda activate paddle_env
# View the current location of python
where python
```
<img src="../install/windows/conda_list_env.png" alt="create environment" width="600" align="center"/>
The above anaconda environment and python environment are installed
<a name="1.2"></a>
### 1.2 Mac
#### 1.2.1 Installing Anaconda
- Note: To use paddlepaddle you need to install the python environment first, here we choose the python integrated environment Anaconda toolkit
- Anaconda is a common python package manager
- After installing Anaconda, you can install the python environment, as well as numpy and other required toolkit environment
- Anaconda download:.
- Address: https://mirrors.tuna.tsinghua.edu.cn/anaconda/archive/?C=M&O=D
<img src="../install/mac/anaconda_start.png" alt="anaconda download" width="800" align="center"/>
- Select `Anaconda3-2021.05-MacOSX-x86_64.pkg` at the bottom to download
- After downloading, double click on the .pkg file to enter the graphical interface
- Just follow the default settings, it will take a while to install
- It is recommended to install a code editor such as vscode or pycharm
#### 1.2.2 Open a terminal and create a conda environment
- Open the terminal
- Press command and spacebar at the same time, type "terminal" in the focus search, double click to enter terminal
- **Add conda to the environment variables**
- Environment variables are added so that the system can recognize the conda command
- Open `~/.bash_profile` in the terminal by typing the following command.
```shell
vim ~/.bash_profile
```
- Add conda as an environment variable in `~/.bash_profile`.
```shell
# Press i first to enter edit mode
# In the first line type.
export PATH="~/opt/anaconda3/bin:$PATH"
# If you customized the installation location during installation, change ~/opt/anaconda3/bin to the bin folder in the customized installation directory
```
```shell
# The modified ~/.bash_profile file should look like this (where xxx is the username)
export PATH="~/opt/anaconda3/bin:$PATH"
# >>> conda initialize >>>
# !!! Contents within this block are managed by 'conda init' !!!
__conda_setup="$('/Users/xxx/opt/anaconda3/bin/conda' 'shell.bash' 'hook' 2> /dev/null)"
if [ $? -eq 0 ]; then
eval "$__conda_setup"
else
if [ -f "/Users/xxx/opt/anaconda3/etc/profile.d/conda.sh" ]; then
. "/Users/xxx/opt/anaconda3/etc/profile.d/conda.sh"
else
export PATH="/Users/xxx/opt/anaconda3/bin:$PATH"
fi
fi
unset __conda_setup
# <<< conda initialize <<<
```
- When you are done, press `esc` to exit edit mode, then type `:wq!` and enter to save and exit
- Verify that the conda command is recognized.
- Enter `source ~/.bash_profile` in the terminal to update the environment variables
- Enter `conda info --envs` in the terminal again, if it shows that there is a base environment, then conda has been added to the environment variables
- Create a new conda environment
```shell
# Enter the following command at the command line to create an environment called paddle_env
# Here to speed up the download, use Tsinghua source
conda create --name paddle_env python=3.8 --channel https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
```
- This command will create an executable environment named paddle_env with python version 3.8, which will take a while depending on the network status
- The command line will then output a prompt, type y and enter to continue the installation
- <img src="../install/mac/conda_create.png" alt="conda_create" width="600" align="center"/>
- To activate the conda environment you just created, enter the following command at the command line.
```shell
# Activate the paddle_env environment
conda activate paddle_env
# View the current location of python
where python
```
<img src="../install/mac/conda_activate.png" alt="conda_actviate" width="600" align="center"/>
The above anaconda environment and python environment are installed
<a name="1.3"></a>
### 1.3 Linux
Linux users can choose to run either Anaconda or Docker. If you are familiar with Docker and need to train the PaddleOCR model, it is recommended to use the Docker environment, where the development process of PaddleOCR is run. If you are not familiar with Docker, you can also use Anaconda to run the project.
#### 1.3.1 Anaconda environment configuration
- Note: To use paddlepaddle you need to install the python environment first, here we choose the python integrated environment Anaconda toolkit
- Anaconda is a common python package manager
- After installing Anaconda, you can install the python environment, as well as numpy and other required toolkit environment
- **Download Anaconda**.
- Download at: https://mirrors.tuna.tsinghua.edu.cn/anaconda/archive/?C=M&O=D
<img src="../install/linux/anaconda_download.png" akt="anaconda download" width="800" align="center"/>
- Select the appropriate version for your operating system
- Type `uname -m` in the terminal to check the command set used by your system
- Download method 1: Download locally, then transfer the installation package to the linux server
- Download method 2: Directly use linux command line to download
```shell
# First install wget
sudo apt-get install wget # Ubuntu
sudo yum install wget # CentOS
```
```bash
# Then use wget to download from Tsinghua source
# If you want to download Anaconda3-2021.05-Linux-x86_64.sh, the download command is as follows
wget https://mirrors.tuna.tsinghua.edu.cn/anaconda/archive/Anaconda3-2021.05-Linux-x86_64.sh
# If you want to download another version, you need to change the file name after the last 1 / to the version you want to download
```
- To install Anaconda.
- Type `sh Anaconda3-2021.05-Linux-x86_64.sh` at the command line
- If you downloaded a different version, replace the file name of the command with the name of the file you downloaded
- Just follow the installation instructions
- You can exit by typing q when viewing the license
- **Add conda to the environment variables**
- If you have already added conda to the environment variable path during the installation, you can skip this step
- Open `~/.bashrc` in a terminal.
```shell
# Enter the following command in the terminal.
vim ~/.bashrc
```
- Add conda as an environment variable in `~/.bashrc`.
```shell
# Press i first to enter edit mode # In the first line enter.
export PATH="~/anaconda3/bin:$PATH"
# If you customized the installation location during installation, change ~/anaconda3/bin to the bin folder in the customized installation directory
```
```shell
# The modified ~/.bash_profile file should look like this (where xxx is the username)
export PATH="~/opt/anaconda3/bin:$PATH"
# >>> conda initialize >>>
# !!! Contents within this block are managed by 'conda init' !!!
__conda_setup="$('/Users/xxx/opt/anaconda3/bin/conda' 'shell.bash' 'hook' 2> /dev/null)"
if [ $? -eq 0 ]; then
eval "$__conda_setup"
else
if [ -f "/Users/xxx/opt/anaconda3/etc/profile.d/conda.sh" ]; then
. "/Users/xxx/opt/anaconda3/etc/profile.d/conda.sh"
else
export PATH="/Users/xxx/opt/anaconda3/bin:$PATH"
fi
fi
unset __conda_setup
# <<< conda initialize <<<
```
- When you are done, press `esc` to exit edit mode, then type `:wq!` and enter to save and exit
- Verify that the conda command is recognized.
- Enter `source ~/.bash_profile` in the terminal to update the environment variables
- Enter `conda info --envs` in the terminal again, if it shows that there is a base environment, then conda has been added to the environment variables
- Create a new conda environment
```shell
# Enter the following command at the command line to create an environment called paddle_env
# Here to speed up the download, use Tsinghua source
conda create --name paddle_env python=3.8 --channel https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
```
- This command will create an executable environment named paddle_env with python version 3.8, which will take a while depending on the network status
- The command line will then output a prompt, type y and enter to continue the installation
<img src="../install/linux/conda_create.png" alt="conda_create" width="500" align="center"/>
- To activate the conda environment you just created, enter the following command at the command line.
```shell
# Activate the paddle_env environment
conda activate paddle_env
```
The above anaconda environment and python environment are installed
#### 1.3.2 Docker environment preparation
**The first time you use this docker image, it will be downloaded automatically. Please be patient.**
```bash
# Switch to the working directory
cd /home/Projects
# You need to create a docker container for the first run, and do not need to run the current command when you run it again
# Create a docker container named ppocr and map the current directory to the /paddle directory of the container
# If using CPU, use docker instead of nvidia-docker to create docker
sudo docker run --name ppocr -v $PWD:/paddle --network=host -it paddlepaddle/paddle:latest-dev-cuda10.1-cudnn7-gcc82 /bin/bash
```
<a name="2"></a>
## 2. Install PaddlePaddle 2.0
- If you have cuda9 or cuda10 installed on your machine, please run the following command to install
```bash
python3 -m pip install paddlepaddle-gpu -i https://mirror.baidu.com/pypi/simple
```
- If you only have cpu on your machine, please run the following command to install
```bash
python3 -m pip install paddlepaddle -i https://mirror.baidu.com/pypi/simple
```
For more software version requirements, please refer to the instructions in [Installation Document](https://www.paddlepaddle.org.cn/install/quick) for operation.
doc/doc_en/inference_ppocr_en.md 0 → 100755
浏览文件 @ a8a9b2e5
# Python Inference for PP-OCR Model Library
This article introduces the use of the Python inference engine for the PP-OCR model library. The content is in order of text detection, text recognition, direction classifier and the prediction method of the three in series on the CPU and GPU.
- [Text Detection Model Inference](#DETECTION_MODEL_INFERENCE)
- [Text Recognition Model Inference](#RECOGNITION_MODEL_INFERENCE)
- [1. Lightweight Chinese Recognition Model Inference](#LIGHTWEIGHT_RECOGNITION)
- [2. Multilingaul Model Inference](#MULTILINGUAL_MODEL_INFERENCE)
- [Angle Classification Model Inference](#ANGLE_CLASS_MODEL_INFERENCE)
- [Text Detection Angle Classification and Recognition Inference Concatenation](#CONCATENATION)
<a name="DETECTION_MODEL_INFERENCE"></a>
## Text Detection Model Inference
The default configuration is based on the inference setting of the DB text detection model. For lightweight Chinese detection model inference, you can execute the following commands:
```
# download DB text detection inference model
wget https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_det_infer.tar
tar xf ch_ppocr_mobile_v2.0_det_infer.tar
# predict
python3 tools/infer/predict_det.py --image_dir="./doc/imgs/00018069.jpg" --det_model_dir="./inference/det_db/"
```
The visual text detection results are saved to the ./inference_results folder by default, and the name of the result file is prefixed with'det_res'. Examples of results are as follows:
![](../imgs_results/det_res_00018069.jpg)
You can use the parameters `limit_type` and `det_limit_side_len` to limit the size of the input image,
The optional parameters of `limit_type` are [`max`, `min`], and
`det_limit_size_len` is a positive integer, generally set to a multiple of 32, such as 960.
The default setting of the parameters is `limit_type='max', det_limit_side_len=960`. Indicates that the longest side of the network input image cannot exceed 960,
If this value is exceeded, the image will be resized with the same width ratio to ensure that the longest side is `det_limit_side_len`.
Set as `limit_type='min', det_limit_side_len=960`, it means that the shortest side of the image is limited to 960.
If the resolution of the input picture is relatively large and you want to use a larger resolution prediction, you can set det_limit_side_len to the desired value, such as 1216:
```
python3 tools/infer/predict_det.py --image_dir="./doc/imgs/1.jpg" --det_model_dir="./inference/det_db/" --det_limit_type=max --det_limit_side_len=1216
```
If you want to use the CPU for prediction, execute the command as follows
```
python3 tools/infer/predict_det.py --image_dir="./doc/imgs/1.jpg" --det_model_dir="./inference/det_db/" --use_gpu=False
```
<a name="RECOGNITION_MODEL_INFERENCE"></a>
## Text Recognition Model Inference
<a name="LIGHTWEIGHT_RECOGNITION"></a>
### 1. Lightweight Chinese Recognition Model Inference
For lightweight Chinese recognition model inference, you can execute the following commands:
```
# download CRNN text recognition inference model
wget https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_infer.tar
tar xf ch_ppocr_mobile_v2.0_rec_infer.tar
python3 tools/infer/predict_rec.py --image_dir="./doc/imgs_words_en/word_10.png" --rec_model_dir="ch_ppocr_mobile_v2.0_rec_infer"
```
![](../imgs_words_en/word_10.png)
After executing the command, the prediction results (recognized text and score) of the above image will be printed on the screen.
```bash
Predicts of ./doc/imgs_words_en/word_10.png:('PAIN', 0.9897658)
```
<a name="MULTILINGUAL_MODEL_INFERENCE"></a>
### 2. Multilingaul Model Inference
If you need to predict other language models, when using inference model prediction, you need to specify the dictionary path used by `--rec_char_dict_path`. At the same time, in order to get the correct visualization results,
You need to specify the visual font path through `--vis_font_path`. There are small language fonts provided by default under the `doc/fonts` path, such as Korean recognition:
```
python3 tools/infer/predict_rec.py --image_dir="./doc/imgs_words/korean/1.jpg" --rec_model_dir="./your inference model" --rec_char_type="korean" --rec_char_dict_path="ppocr/utils/dict/korean_dict.txt" --vis_font_path="doc/fonts/korean.ttf"
```
![](../imgs_words/korean/1.jpg)
After executing the command, the prediction result of the above figure is:
``` text
Predicts of ./doc/imgs_words/korean/1.jpg:('바탕으로', 0.9948904)
```
<a name="ANGLE_CLASS_MODEL_INFERENCE"></a>
## Angle Classification Model Inference
For angle classification model inference, you can execute the following commands:
```
# download text angle class inference model：
wget https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_infer.tar
tar xf ch_ppocr_mobile_v2.0_cls_infer.tar
python3 tools/infer/predict_cls.py --image_dir="./doc/imgs_words_en/word_10.png" --cls_model_dir="ch_ppocr_mobile_v2.0_cls_infer"
```
![](../imgs_words_en/word_10.png)
After executing the command, the prediction results (classification angle and score) of the above image will be printed on the screen.
```
Predicts of ./doc/imgs_words_en/word_10.png:['0', 0.9999995]
```
<a name="CONCATENATION"></a>
## Text Detection Angle Classification and Recognition Inference Concatenation
When performing prediction, you need to specify the path of a single image or a folder of images through the parameter `image_dir`, the parameter `det_model_dir` specifies the path to detect the inference model, the parameter `cls_model_dir` specifies the path to angle classification inference model and the parameter `rec_model_dir` specifies the path to identify the inference model. The parameter `use_angle_cls` is used to control whether to enable the angle classification model. The parameter `use_mp` specifies whether to use multi-process to infer `total_process_num` specifies process number when using multi-process. The parameter . The visualized recognition results are saved to the `./inference_results` folder by default.
```shell
# use direction classifier
python3 tools/infer/predict_system.py --image_dir="./doc/imgs/00018069.jpg" --det_model_dir="./inference/det_db/" --cls_model_dir="./inference/cls/" --rec_model_dir="./inference/rec_crnn/" --use_angle_cls=true
# not use use direction classifier
python3 tools/infer/predict_system.py --image_dir="./doc/imgs/00018069.jpg" --det_model_dir="./inference/det_db/" --rec_model_dir="./inference/rec_crnn/"
# use multi-process
python3 tools/infer/predict_system.py --image_dir="./doc/imgs/00018069.jpg" --det_model_dir="./inference/det_db/" --rec_model_dir="./inference/rec_crnn/" --use_angle_cls=false --use_mp=True --total_process_num=6
```
After executing the command, the recognition result image is as follows:
![](../imgs_results/system_res_00018069.jpg)
doc/doc_en/models_and_config_en.md 0 → 100644
浏览文件 @ a8a9b2e5
# PP-OCR Model and Configuration
The chapter on PP-OCR model and configuration file mainly adds some basic concepts of OCR model and the content and role of configuration file to have a better experience in the subsequent parameter adjustment and training of the model.
This chapter contains three parts. Firstly, [PP-OCR Model Download](. /models_list_en.md) explains the concept of PP-OCR model types and provides links to download all models. Then in [Yml Configuration](. /config_en.md) details the parameters needed to fine-tune the PP-OCR models. The final [Python Inference for PP-OCR Model Library](. /inference_ppocr_en.md) is an introduction to the use of the PP-OCR model library in the first section, which can quickly utilize the rich model library models to obtain test results through the Python inference engine.
------
Let's first understand some basic concepts.
- [INTRODUCTION ABOUT OCR](#introduction-about-ocr)
* [BASIC CONCEPTS OF OCR DETECTION MODEL](#basic-concepts-of-ocr-detection-model)
* [Basic concepts of OCR recognition model](#basic-concepts-of-ocr-recognition-model)
* [PP-OCR model](#pp-ocr-model)
* [And a table of contents](#and-a-table-of-contents)
* [On the right](#on-the-right)
## 1. INTRODUCTION ABOUT OCR
This section briefly introduces the basic concepts of OCR detection model and recognition model, and introduces PaddleOCR's PP-OCR model.
OCR (Optical Character Recognition, Optical Character Recognition) is currently the general term for text recognition. It is not limited to document or book text recognition, but also includes recognizing text in natural scenes. It can also be called STR (Scene Text Recognition).
OCR text recognition generally includes two parts, text detection and text recognition. The text detection module first uses detection algorithms to detect text lines in the image. And then the recognition algorithm to identify the specific text in the text line.
### 1.1 BASIC CONCEPTS OF OCR DETECTION MODEL
Text detection can locate the text area in the image, and then usually mark the word or text line in the form of a bounding box. Traditional text detection algorithms mostly extract features manually, which are characterized by fast speed and good effect in simple scenes, but the effect will be greatly reduced when faced with natural scenes. Currently, deep learning methods are mostly used.
Text detection algorithms based on deep learning can be roughly divided into the following categories:
1. Method based on target detection. Generally, after the text box is predicted, the final text box is filtered through NMS, which is mostly four-point text box, which is not ideal for curved text scenes. Typical algorithms are methods such as EAST and Text Box.
2. Method based on text segmentation. The text line is regarded as the segmentation target, and then the external text box is constructed through the segmentation result, which can handle curved text, and the effect is not ideal for the text cross scene problem. Typical algorithms are DB, PSENet and other methods.
3. Hybrid target detection and segmentation method.
### 1.2 Basic concepts of OCR recognition model
The input of the OCR recognition algorithm is generally text lines images which has less background information, and the text information occupies the main part. The recognition algorithm can be divided into two types of algorithms:
1. CTC-based method. The text prediction module of the recognition algorithm is based on CTC, and the commonly used algorithm combination is CNN+RNN+CTC. There are also some algorithms that try to add transformer modules to the network and so on.
2. Attention-based method. The text prediction module of the recognition algorithm is based on Attention, and the commonly used algorithm combination is CNN+RNN+Attention.
### 1.3 PP-OCR model
PaddleOCR integrates many OCR algorithms, text detection algorithms include DB, EAST, SAST, etc., text recognition algorithms include CRNN, RARE, StarNet, Rosetta, SRN and other algorithms.
Among them, PaddleOCR has released the PP-OCR series model for the general OCR in Chinese and English natural scenes. The PP-OCR model is composed of the DB+CRNN algorithm. It uses massive Chinese data training and model tuning methods to have high text detection and recognition capabilities in Chinese scenes. And PaddleOCR has launched a high-precision and ultra-lightweight PP-OCRv2 model. The detection model is only 3M, and the recognition model is only 8.5M. Using [PaddleSlim](https://github.com/PaddlePaddle/PaddleSlim)'s model quantification method, the detection model can be compressed to 0.8M without reducing the accuracy. The recognition is compressed to 3M, which is more suitable for mobile deployment scenarios.
doc/doc_en/models_list_en.md
浏览文件 @ a8a9b2e5
## OCR model list（V2.0, updated on 2021.1.20）
## OCR model list（V2.1, updated on 2021.9.6）
> **Note**
> 1. Compared with [models 1.1](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/doc/doc_en/models_list_en.md), which are trained with static graph programming paradigm, models 2.0 are the dynamic graph trained version and achieve close performance.
> 2. All models in this tutorial are all ppocr-series models, for more introduction of algorithms and models based on public dataset, you can refer to [algorithm overview tutorial](./algorithm_overview_en.md).
> 1. Compared with the model v2.0, the 2.1 version of the detection model has a improvement in accuracy, and the 2.1 version of the recognition model is optimized in accuracy and CPU speed.
> 2. Compared with [models 1.1](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/doc/doc_en/models_list_en.md), which are trained with static graph programming paradigm, models 2.0 are the dynamic graph trained version and achieve close performance.
> 3. All models in this tutorial are all ppocr-series models, for more introduction of algorithms and models based on public dataset, you can refer to [algorithm overview tutorial](./algorithm_overview_en.md).
- [1. Text Detection Model](#Detection)
- [2. Text Recognition Model](#Recognition)
......@@ -28,6 +29,8 @@ Relationship of the above models is as follows.
|model name|description|config|model size|download|
| --- | --- | --- | --- | --- |
|ch_PP-OCRv2_det_slim|slim quantization with distillation lightweight model, supporting Chinese, English, multilingual text detection|[ch_PP-OCRv2_det_cml.yml](../../configs/det/ch_PP-OCRv2/ch_PP-OCR_det_cml.yml)| 3M |[inference model](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_det_slim_quant_infer.tar)|
|ch_PP-OCRv2_det|Original lightweight model, supporting Chinese, English, multilingual text detection|[ch_PP-OCRv2_det_cml.yml](../../configs/det/ch_PP-OCRv2/ch_PP-OCR_det_cml.yml)|3M|[inference model](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_det_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_det_distill_train.tar)|
|ch_ppocr_mobile_slim_v2.0_det|Slim pruned lightweight model, supporting Chinese, English, multilingual text detection|[ch_det_mv3_db_v2.0.yml](../../configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml)|2.6M |[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/slim/ch_ppocr_mobile_v2.0_det_prune_infer.tar)|
|ch_ppocr_mobile_v2.0_det|Original lightweight model, supporting Chinese, English, multilingual text detection|[ch_det_mv3_db_v2.0.yml](../../configs/det/ch_ppocr_v2.0/ch_det_mv3_db_v2.0.yml)|3M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_det_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_det_train.tar)|
|ch_ppocr_server_v2.0_det|General model, which is larger than the lightweight model, but achieved better performance|[ch_det_res18_db_v2.0.yml](../../configs/det/ch_ppocr_v2.0/ch_det_res18_db_v2.0.yml)|47M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_det_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_det_train.tar)|
......@@ -40,6 +43,8 @@ Relationship of the above models is as follows.
|model name|description|config|model size|download|
| --- | --- | --- | --- | --- |
|ch_PP-OCRv2_rec_slim|Slim qunatization with distillation lightweight model, supporting Chinese, English, multilingual text detection|[ch_PP-OCRv2_rec.yml](../../configs/rec/ch_PP-OCRv2/ch_PP-OCRv2_rec.yml)| 9M |[inference model](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_rec_slim_quant_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_rec_slim_quant_train.tar) |
|ch_PP-OCRv2_rec|Original lightweight model, supporting Chinese, English, multilingual text detection|[ch_PP-OCRv2_rec.yml](../../configs/rec/ch_PP-OCRv2/ch_PP-OCRv2_rec.yml)|8.5M|[inference model](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_rec_train.tar) |
|ch_ppocr_mobile_slim_v2.0_rec|Slim pruned and quantized lightweight model, supporting Chinese, English and number recognition|[rec_chinese_lite_train_v2.0.yml](../../configs/rec/ch_ppocr_v2.0/rec_chinese_lite_train_v2.0.yml)| 6M | [inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_slim_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_slim_train.tar) |
|ch_ppocr_mobile_v2.0_rec|Original lightweight model, supporting Chinese, English and number recognition|[rec_chinese_lite_train_v2.0.yml](../../configs/rec/ch_ppocr_v2.0/rec_chinese_lite_train_v2.0.yml)|5.2M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_train.tar) / [pre-trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_pre.tar) |
|ch_ppocr_server_v2.0_rec|General model, supporting Chinese, English and number recognition|[rec_chinese_common_train_v2.0.yml](../../configs/rec/ch_ppocr_v2.0/rec_chinese_common_train_v2.0.yml)|94.8M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_rec_train.tar) / [pre-trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_rec_pre.tar) |
......@@ -58,45 +63,6 @@ Relationship of the above models is as follows.
<a name="Multilingual"></a>
#### Multilingual Recognition Model（Updating...）
**Note：** The configuration file of the new multi language model is generated by code. You can use the `--help` parameter to check which multi language are supported by current PaddleOCR.
```bash
# The code needs to run in the specified directory
cd {your/path/}PaddleOCR/configs/rec/multi_language/
python3 generate_multi_language_configs.py --help
```
Take the Italian configuration file as an example：
##### 1.Generate Italian configuration file to test the model provided
you can generate the default configuration file through the following command, and use the default language dictionary provided by paddleocr for prediction.
```bash
# The code needs to run in the specified directory
cd {your/path/}PaddleOCR/configs/rec/multi_language/
# Set the required language configuration file through -l or --language parameter
# This command will write the default parameter to the configuration file.
python3 generate_multi_language_configs.py -l it
```
##### 2. Generate Italian configuration file to train your own data
If you want to train your own model, you can prepare the training set file, verification set file, dictionary file and training data path. Here we assume that the Italian training set, verification set, dictionary and training data path are:
- Training set:{your/path/}PaddleOCR/train_data/train_list.txt
- Validation set: {your/path/}PaddleOCR/train_data/val_list.txt
- Use the default dictionary provided by paddleocr:{your/path/}PaddleOCR/ppocr/utils/dict/it_dict.txt
- Training data path:{your/path/}PaddleOCR/train_data
```bash
# The code needs to run in the specified directory
cd {your/path/}PaddleOCR/configs/rec/multi_language/
# The -l or --language parameter is required
# --train modify train_list path
# --val modify eval_list path
# --data_dir modify data dir
# -o modify default parameters
# --dict Change the dictionary path. The example uses the default dictionary path, so that this parameter can be empty.
python3 generate_multi_language_configs.py -l it \
--train {path/to/train_list} \
--val {path/to/val_list} \
--data_dir {path/to/data_dir} \
-o Global.use_gpu=False
```
|model name| dict file | description|config|model size|download|
| --- | --- | --- |--- | --- | --- |
| french_mobile_v2.0_rec | ppocr/utils/dict/french_dict.txt | Lightweight model for French recognition|[rec_french_lite_train.yml](../../configs/rec/multi_language/rec_french_lite_train.yml)|2.65M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/french_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/french_mobile_v2.0_rec_train.tar) |
......@@ -120,12 +86,14 @@ For more supported languages, please refer to : [Multi-language model](./multi_l
|model name|description|config|model size|download|
| --- | --- | --- | --- | --- |
|ch_ppocr_mobile_slim_v2.0_cls|Slim quantized model|[cls_mv3.yml](../../configs/cls/cls_mv3.yml)| 2.1M | [inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_slim_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_slim_train.tar) |
|ch_ppocr_mobile_v2.0_cls|Original model|[cls_mv3.yml](../../configs/cls/cls_mv3.yml)|1.38M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_train.tar) |
|ch_ppocr_mobile_slim_v2.0_cls|Slim quantized model for text angle classification|[cls_mv3.yml](../../configs/cls/cls_mv3.yml)| 2.1M | [inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_slim_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_slim_train.tar) |
|ch_ppocr_mobile_v2.0_cls|Original model for text angle classification|[cls_mv3.yml](../../configs/cls/cls_mv3.yml)|1.38M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_cls_train.tar) |
<a name="Paddle-Lite"></a>
### 4. Paddle-Lite Model
|Version|Introduction|Model size|Detection model|Text Direction model|Recognition model|Paddle-Lite branch|
|---|---|---|---|---|---|---|
|V2.0|extra-lightweight chinese OCR optimized model|7.8M|[download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_det_opt.nb)|[download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_cls_opt.nb)|[download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_rec_opt.nb)|v2.9|
|V2.0(slim)|extra-lightweight chinese OCR optimized model|3.3M|[download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_det_slim_opt.nb)|[download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_cls_slim_opt.nb)|[download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_rec_slim_opt.nb)|v2.9|
|PP-OCRv2|extra-lightweight chinese OCR optimized model|11M|[download link](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_det_infer_opt.nb)|[download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_cls_opt.nb)|[download link](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_rec_infer_opt.nb)|v2.9|
|PP-OCRv2(slim)|extra-lightweight chinese OCR optimized model|4.9M|[download link](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_det_slim_opt.nb)|[download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_cls_slim_opt.nb)|[download link](https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_rec_slim_opt.nb)|v2.9|
|V2.0|ppocr_v2.0 extra-lightweight chinese OCR optimized model|7.8M|[download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_det_opt.nb)|[download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_cls_opt.nb)|[download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_rec_opt.nb)|v2.9|
|V2.0(slim)|ppovr_v2.0 extra-lightweight chinese OCR optimized model|3.3M|[download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_det_slim_opt.nb)|[download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_cls_slim_opt.nb)|[download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/lite/ch_ppocr_mobile_v2.0_rec_slim_opt.nb)|v2.9|
doc/doc_en/multi_languages_en.md
浏览文件 @ a8a9b2e5
此差异已折叠。
doc/doc_en/paddleOCR_overview_en.md 0 → 100644
浏览文件 @ a8a9b2e5
此差异已折叠。
doc/doc_en/quickstart_en.md
浏览文件 @ a8a9b2e5
此差异已折叠。
doc/doc_en/recognition_en.md
浏览文件 @ a8a9b2e5
此差异已折叠。
doc/doc_en/training_en.md 0 → 100644
浏览文件 @ a8a9b2e5
此差异已折叠。
doc/doc_en/update_en.md
浏览文件 @ a8a9b2e5
# RECENT UPDATES
- 2021.9.7 release PaddleOCR v2.3, [PP-OCRv2](#PP-OCRv2) is proposed. The CPU inference speed of PP-OCRv2 is 220% higher than that of PP-OCR server. The F-score of PP-OCRv2 is 7% higher than that of PP-OCR mobile.
- 2021.8.3 released PaddleOCR v2.2, add a new structured documents analysis toolkit, i.e., [PP-Structure](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.2/ppstructure/README.md), support layout analysis and table recognition (One-key to export chart images to Excel files).
- 2021.4.8 release end-to-end text recognition algorithm [PGNet](https://www.aaai.org/AAAI21Papers/AAAI-2885.WangP.pdf) which is published in AAAI 2021. Find tutorial [here](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.1/doc/doc_en/pgnet_en.md)；release multi language recognition [models](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.1/doc/doc_en/multi_languages_en.md), support more than 80 languages recognition; especically, the performance of [English recognition model](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.1/doc/doc_en/models_list_en.md#English) is Optimized.
- 2021.1.21 update more than 25+ multilingual recognition models [models list](./doc/doc_en/models_list_en.md), including：English, Chinese, German, French, Japanese，Spanish，Portuguese Russia Arabic and so on. Models for more languages will continue to be updated [Develop Plan](https://github.com/PaddlePaddle/PaddleOCR/issues/1048).
- 2020.12.15 update Data synthesis tool, i.e., [Style-Text](../../StyleText/README.md)，easy to synthesize a large number of images which are similar to the target scene image.
- 2020.11.25 Update a new data annotation tool, i.e., [PPOCRLabel](../../PPOCRLabel/README.md), which is helpful to improve the labeling efficiency. Moreover, the labeling results can be used in training of the PP-OCR system directly.
- 2020.9.22 Update the PP-OCR technical article, https://arxiv.org/abs/2009.09941
......
doc/doc_en/visualization_en.md
浏览文件 @ a8a9b2e5
此差异已折叠。
doc/doc_en/whl_en.md
浏览文件 @ a8a9b2e5
# paddleocr package
# Paddleocr Package
## 1 Get started quickly
### 1.1 install package
......
doc/imgs_results/PP-OCRv2/PP-OCRv2-pic002.jpg 0 → 100644
浏览文件 @ a8a9b2e5
此差异已折叠。
doc/imgs_results/PP-OCRv2/PP-OCRv2-pic003.jpg 0 → 100644
浏览文件 @ a8a9b2e5
此差异已折叠。
doc/install/linux/anaconda_download.png 0 → 100755
浏览文件 @ a8a9b2e5
此差异已折叠。
doc/install/linux/conda_create.png 0 → 100755
浏览文件 @ a8a9b2e5
此差异已折叠。
doc/install/mac/anaconda_start.png 0 → 100755
浏览文件 @ a8a9b2e5
此差异已折叠。
doc/install/mac/conda_activate.png 0 → 100755
浏览文件 @ a8a9b2e5
此差异已折叠。
doc/install/mac/conda_create.png 0 → 100755
浏览文件 @ a8a9b2e5
此差异已折叠。
doc/install/windows/Anaconda_download.png 0 → 100644
浏览文件 @ a8a9b2e5
此差异已折叠。
doc/install/windows/anaconda_install_env.png 0 → 100644
浏览文件 @ a8a9b2e5
此差异已折叠。
doc/install/windows/anaconda_install_folder.png 0 → 100644
浏览文件 @ a8a9b2e5
此差异已折叠。
doc/install/windows/anaconda_prompt.png 0 → 100755
浏览文件 @ a8a9b2e5
此差异已折叠。
doc/install/windows/conda_list_env.png 0 → 100644
浏览文件 @ a8a9b2e5
此差异已折叠。
doc/install/windows/conda_new_env.png 0 → 100644
浏览文件 @ a8a9b2e5
此差异已折叠。
doc/joinus.PNG
浏览文件 @ a8a9b2e5
此差异已折叠。
doc/overview.png 0 → 100644
浏览文件 @ a8a9b2e5
此差异已折叠。
doc/overview_en.png 0 → 100644
浏览文件 @ a8a9b2e5
此差异已折叠。
doc/ppocrv2_framework.jpg 0 → 100644
浏览文件 @ a8a9b2e5
此差异已折叠。
doc/table/1.png
浏览文件 @ a8a9b2e5
此差异已折叠。
doc/table/table.jpg
浏览文件 @ a8a9b2e5
此差异已折叠。
paddleocr.py
浏览文件 @ a8a9b2e5
此差异已折叠。
ppocr/data/__init__.py
浏览文件 @ a8a9b2e5
此差异已折叠。
ppocr/data/imaug/__init__.py
浏览文件 @ a8a9b2e5
此差异已折叠。
ppocr/data/imaug/label_ops.py
浏览文件 @ a8a9b2e5
此差异已折叠。
ppocr/data/imaug/operators.py
浏览文件 @ a8a9b2e5
此差异已折叠。
ppocr/data/imaug/rec_img_aug.py
浏览文件 @ a8a9b2e5
此差异已折叠。
ppocr/losses/__init__.py
浏览文件 @ a8a9b2e5
此差异已折叠。
ppocr/losses/cls_loss.py
浏览文件 @ a8a9b2e5
此差异已折叠。
ppocr/losses/rec_nrtr_loss.py 0 → 100644
浏览文件 @ a8a9b2e5
此差异已折叠。
ppocr/metrics/rec_metric.py
浏览文件 @ a8a9b2e5
此差异已折叠。
ppocr/modeling/backbones/rec_nrtr_mtb.py 0 → 100644
浏览文件 @ a8a9b2e5
此差异已折叠。
ppocr/modeling/heads/__init__.py
浏览文件 @ a8a9b2e5
此差异已折叠。
ppocr/modeling/heads/multiheadAttention.py 0 → 100755
浏览文件 @ a8a9b2e5
此差异已折叠。
ppocr/modeling/heads/rec_nrtr_head.py 0 → 100644
浏览文件 @ a8a9b2e5
此差异已折叠。
ppocr/postprocess/__init__.py
浏览文件 @ a8a9b2e5
此差异已折叠。
ppstructure/README.md
浏览文件 @ a8a9b2e5
此差异已折叠。
ppstructure/README_ch.md
浏览文件 @ a8a9b2e5
此差异已折叠。
ppstructure/table/README.md
浏览文件 @ a8a9b2e5
此差异已折叠。
ppstructure/table/README_ch.md
浏览文件 @ a8a9b2e5
此差异已折叠。
requirements.txt
浏览文件 @ a8a9b2e5
此差异已折叠。
tests/compare_results.py 0 → 100644
浏览文件 @ a8a9b2e5
此差异已折叠。
tests/configs/det_mv3_db.yml 0 → 100644
浏览文件 @ a8a9b2e5
此差异已折叠。
tests/configs/det_r50_vd_db.yml 0 → 100644
浏览文件 @ a8a9b2e5
此差异已折叠。
tests/ocr_det_params.txt
浏览文件 @ a8a9b2e5
此差异已折叠。
tests/ocr_det_server_params.txt 0 → 100644
浏览文件 @ a8a9b2e5
此差异已折叠。
tests/ocr_ppocr_mobile_params.txt 0 → 100644
浏览文件 @ a8a9b2e5
此差异已折叠。
tests/prepare.sh
浏览文件 @ a8a9b2e5
此差异已折叠。
tests/readme.md 0 → 100644
浏览文件 @ a8a9b2e5
此差异已折叠。
tests/results/det_results_gpu_fp32.txt 0 → 100644
浏览文件 @ a8a9b2e5
此差异已折叠。
tests/results/det_results_gpu_trt_fp16.txt 0 → 100644
浏览文件 @ a8a9b2e5
此差异已折叠。
tests/test.sh
浏览文件 @ a8a9b2e5
此差异已折叠。
tools/eval.py
浏览文件 @ a8a9b2e5
此差异已折叠。
tools/export_model.py
浏览文件 @ a8a9b2e5
此差异已折叠。
tools/infer/predict_det.py
浏览文件 @ a8a9b2e5
此差异已折叠。
tools/infer/predict_e2e.py
浏览文件 @ a8a9b2e5
此差异已折叠。
tools/infer/predict_rec.py
浏览文件 @ a8a9b2e5
此差异已折叠。
tools/infer/predict_system.py
浏览文件 @ a8a9b2e5
此差异已折叠。
tools/infer/utility.py
浏览文件 @ a8a9b2e5
此差异已折叠。
tools/infer_det.py
浏览文件 @ a8a9b2e5
此差异已折叠。
tools/infer_rec.py
浏览文件 @ a8a9b2e5
此差异已折叠。
tools/program.py
浏览文件 @ a8a9b2e5
此差异已折叠。
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或