Skip to content

P
PaddleOCR

PaddlePaddle / PaddleOCR
大约 2 年 前同步成功

通知 1557
Star 32965
Fork 6643
P
PaddleOCR
“e976fcb91d7be2a7e48485f752975a851c1506e6”上不存在“source/libs/catalog/git@gitcode.net:taosdata/tdengine.git”
提交 5e9fb50d 编写于 作者: T tink2123
浏览文件
操作

Merge branch 'dygraph' of https://github.com/PaddlePaddle/PaddleOCR into multi_languages

上级 45117f90 5a5d627d
显示空白变更内容
内联 并排
Showing with 334 addition and 207 deletion
PPOCRLabel/PPOCRLabel.py
浏览文件 @ 5e9fb50d
......@@ -24,6 +24,7 @@ import sys
from functools import partial
from collections import defaultdict
import json
import cv2
__dir__ = os.path.dirname(os.path.abspath(__file__))
......@@ -1242,10 +1243,13 @@ class MainWindow(QMainWindow, WindowMixin):
# if unicodeFilePath in self.mImgList:
if unicodeFilePath and os.path.exists(unicodeFilePath):
self.imageData = read(unicodeFilePath, None)
self.canvas.verified = False
image = QImage.fromData(self.imageData)
cvimg = cv2.imdecode(np.fromfile(unicodeFilePath, dtype=np.uint8), 1)
height, width, depth = cvimg.shape
cvimg = cv2.cvtColor(cvimg, cv2.COLOR_BGR2RGB)
image = QImage(cvimg.data, width, height, width * depth, QImage.Format_RGB888)
if image.isNull():
self.errorMessage(u'Error opening file',
u"<p>Make sure <i>%s</i> is a valid image file." % unicodeFilePath)
......
PPOCRLabel/libs/autoDialog.py
浏览文件 @ 5e9fb50d
......@@ -7,6 +7,8 @@ except ImportError:
from PyQt4.QtCore import *
import json
import cv2
import numpy as np
from libs.utils import newIcon
......@@ -34,11 +36,16 @@ class Worker(QThread):
if self.handle == 0:
self.listValue.emit(Imgpath)
if self.model == 'paddle':
h, w, _ = cv2.imdecode(np.fromfile(Imgpath, dtype=np.uint8), 1).shape
if h > 32 and w > 32:
self.result_dic = self.ocr.ocr(Imgpath, cls=True, det=True)
else:
print('The size of', Imgpath, 'is too small to be recognised')
self.result_dic = None
# 结果保存
if self.result_dic is None or len(self.result_dic) == 0:
print('Can not recognise file is : ', Imgpath)
print('Can not recognise file', Imgpath)
pass
else:
strs = ''
......
README_ch.md
浏览文件 @ 5e9fb50d
......@@ -8,7 +8,7 @@ PaddleOCR同时支持动态图与静态图两种编程范式
- 静态图版本：develop分支
**近期更新**
- 2021.1.18 [FAQ](./doc/doc_ch/FAQ.md)新增5个高频问题，总数152个，每周一都会更新，欢迎大家持续关注。
- 2021.1.25 [FAQ](./doc/doc_ch/FAQ.md)新增5个高频问题，总数157个，每周一都会更新，欢迎大家持续关注。
- 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
......
deploy/docker/hubserving/cpu/Dockerfile
浏览文件 @ 5e9fb50d
# Version: 1.0.0
FROM hub.baidubce.com/paddlepaddle/paddle:latest-gpu-cuda10.0-cudnn7-dev
# Version: 2.0.0
FROM registry.baidubce.com/paddlepaddle/paddle:2.0.0rc1
# PaddleOCR base on Python3.7
RUN pip3.7 install --upgrade pip -i https://mirror.baidu.com/pypi/simple
RUN python3.7 -m pip install paddlepaddle==2.0.0rc0 -i https://mirror.baidu.com/pypi/simple
RUN pip3.7 install paddlehub --upgrade -i https://mirror.baidu.com/pypi/simple
RUN git clone https://github.com/PaddlePaddle/PaddleOCR.git /PaddleOCR
......@@ -15,15 +13,15 @@ WORKDIR /PaddleOCR
RUN pip3.7 install -r requirements.txt -i https://mirror.baidu.com/pypi/simple
RUN mkdir -p /PaddleOCR/inference/
# Download orc detect model(light version). if you want to change normal version, you can change ch_ppocr_mobile_v1.1_det_infer to ch_ppocr_server_v1.1_det_infer, also remember change det_model_dir in deploy/hubserving/ocr_system/params.py）
# Download orc detect model(light version). if you want to change normal version, you can change ch_ppocr_mobile_v2.0_det_infer to ch_ppocr_server_v2.0_det_infer, also remember change det_model_dir in deploy/hubserving/ocr_system/params.py）
ADD {link} /PaddleOCR/inference/
RUN tar xf /PaddleOCR/inference/{file} -C /PaddleOCR/inference/
# Download direction classifier(light version). If you want to change normal version, you can change ch_ppocr_mobile_v1.1_cls_infer to ch_ppocr_mobile_v1.1_cls_infer, also remember change cls_model_dir in deploy/hubserving/ocr_system/params.py）
# Download direction classifier(light version). If you want to change normal version, you can change ch_ppocr_mobile_v2.0_cls_infer to ch_ppocr_mobile_v2.0_cls_infer, also remember change cls_model_dir in deploy/hubserving/ocr_system/params.py）
ADD {link} /PaddleOCR/inference/
RUN tar xf /PaddleOCR/inference/{file}.tar -C /PaddleOCR/inference/
# Download orc recognition model(light version). If you want to change normal version, you can change ch_ppocr_mobile_v1.1_rec_infer to ch_ppocr_server_v1.1_rec_infer, also remember change rec_model_dir in deploy/hubserving/ocr_system/params.py）
# Download orc recognition model(light version). If you want to change normal version, you can change ch_ppocr_mobile_v2.0_rec_infer to ch_ppocr_server_v2.0_rec_infer, also remember change rec_model_dir in deploy/hubserving/ocr_system/params.py）
ADD {link} /PaddleOCR/inference/
RUN tar xf /PaddleOCR/inference/{file}.tar -C /PaddleOCR/inference/
......
deploy/docker/hubserving/gpu/Dockerfile
浏览文件 @ 5e9fb50d
# Version: 1.0.0
FROM hub.baidubce.com/paddlepaddle/paddle:latest-gpu-cuda10.0-cudnn7-dev
# Version: 2.0.0
FROM egistry.baidubce.com/paddlepaddle/paddle:2.0.0rc1-gpu-cuda10.0-cudnn7
# PaddleOCR base on Python3.7
RUN pip3.7 install --upgrade pip -i https://mirror.baidu.com/pypi/simple
RUN python3.7 -m pip install paddlepaddle-gpu==2.0.0rc0 -i https://mirror.baidu.com/pypi/simple
RUN pip3.7 install paddlehub --upgrade -i https://mirror.baidu.com/pypi/simple
RUN git clone https://github.com/PaddlePaddle/PaddleOCR.git /PaddleOCR
......@@ -15,15 +13,15 @@ WORKDIR /PaddleOCR
RUN pip3.7 install -r requirements.txt -i https://mirror.baidu.com/pypi/simple
RUN mkdir -p /PaddleOCR/inference/
# Download orc detect model(light version). if you want to change normal version, you can change ch_ppocr_mobile_v1.1_det_infer to ch_ppocr_server_v1.1_det_infer, also remember change det_model_dir in deploy/hubserving/ocr_system/params.py）
# Download orc detect model(light version). if you want to change normal version, you can change ch_ppocr_mobile_v2.0_det_infer to ch_ppocr_server_v2.0_det_infer, also remember change det_model_dir in deploy/hubserving/ocr_system/params.py）
ADD {link} /PaddleOCR/inference/
RUN tar xf /PaddleOCR/inference/{file}.tar -C /PaddleOCR/inference/
# Download direction classifier(light version). If you want to change normal version, you can change ch_ppocr_mobile_v1.1_cls_infer to ch_ppocr_mobile_v1.1_cls_infer, also remember change cls_model_dir in deploy/hubserving/ocr_system/params.py）
# Download direction classifier(light version). If you want to change normal version, you can change ch_ppocr_mobile_v2.0_cls_infer to ch_ppocr_mobile_v2.0_cls_infer, also remember change cls_model_dir in deploy/hubserving/ocr_system/params.py）
ADD {link} /PaddleOCR/inference/
RUN tar xf /PaddleOCR/inference/{file} -C /PaddleOCR/inference/
# Download orc recognition model(light version). If you want to change normal version, you can change ch_ppocr_mobile_v1.1_rec_infer to ch_ppocr_server_v1.1_rec_infer, also remember change rec_model_dir in deploy/hubserving/ocr_system/params.py）
# Download orc recognition model(light version). If you want to change normal version, you can change ch_ppocr_mobile_v2.0_rec_infer to ch_ppocr_server_v2.0_rec_infer, also remember change rec_model_dir in deploy/hubserving/ocr_system/params.py）
ADD {link} /PaddleOCR/inference/
RUN tar xf /PaddleOCR/inference/{file}.tar -C /PaddleOCR/inference/
......
doc/doc_ch/FAQ.md
浏览文件 @ 5e9fb50d
......@@ -9,42 +9,43 @@
## PaddleOCR常见问题汇总(持续更新)
* [近期更新（2021.1.18）](#近期更新)
* [近期更新（2021.1.25）](#近期更新)
* [【精选】OCR精选10个问题](#OCR精选10个问题)
* [【理论篇】OCR通用32个问题](#OCR通用问题)
* [基础知识7题](#基础知识)
* [数据集7题](#数据集2)
* [模型训练调优18题](#模型训练调优2)
* [【实战篇】PaddleOCR实战110个问题](#PaddleOCR实战问题)
* [使用咨询36题](#使用咨询)
* [【实战篇】PaddleOCR实战115个问题](#PaddleOCR实战问题)
* [使用咨询38题](#使用咨询)
* [数据集17题](#数据集3)
* [模型训练调优28题](#模型训练调优3)
* [预测部署29题](#预测部署3)
* [预测部署32题](#预测部署3)
<a name="近期更新"></a>
## 近期更新（2021.1.18）
## 近期更新（2021.1.25）
#### Q3.1.37: 小语种模型只有识别模型，没有检测模型吗？
#### Q2.3.18: 在PP-OCR系统中，文本检测的骨干网络为什么没有使用SE模块？
**A**：小语种（包括纯英文数字）的检测模型和中文的检测模型是共用的，在训练中文检测模型时加入了多语言数据。https://github.com/PaddlePaddle/PaddleOCR/blob/dygraph/doc/doc_en/models_list_en.md#1-text-detection-model。
**A**：SE模块是MobileNetV3网络一个重要模块，目的是估计特征图每个特征通道重要性，给特征图每个特征分配权重，提高网络的表达能力。但是，对于文本检测，输入网络的分辨率比较大，一般是640\*640，利用SE模块估计特征图每个特征通道重要性比较困难，网络提升能力有限，但是该模块又比较耗时，因此在PP-OCR系统中，文本检测的骨干网络没有使用SE模块。实验也表明，当去掉SE模块，超轻量模型大小可以减小40%，文本检测效果基本不受影响。详细可以参考PP-OCR技术文章，https://arxiv.org/abs/2009.09941.
#### Q3.1.38: module 'paddle.distributed' has no attribute ‘get_rank’。
#### Q3.3.27: PaddleOCR关于文本识别模型的训练，支持的数据增强方式有哪些？
**A**：Paddle版本问题，请安装2.0版本Paddle：pip install paddlepaddle==2.0.0rc1。
**A**：文本识别支持的数据增强方式有随机小幅度裁剪、图像平衡、添加白噪声、颜色漂移、图像反色和Text Image Augmentation（TIA）变换等。可以参考[代码](../../ppocr/data/imaug/rec_img_aug.py)中的warp函数。
#### Q3.4.30: PaddleOCR是否支持在华为鲲鹏920CPU上部署？
#### Q3.3.28: 关于dygraph分支中，文本识别模型训练，要使用数据增强应该如何设置？
**A**：目前Paddle的预测库是支持华为鲲鹏920CPU的，但是OCR还没在这些芯片上测试过，可以自己调试，有问题反馈给我们。
**A**：可以参考[配置文件](../../configs/rec/ch_ppocr_v2.0/rec_chinese_lite_train_v2.0.yml)在Train['dataset']['transforms']添加RecAug字段，使数据增强生效。可以通过添加对aug_prob设置，表示每种数据增强采用的概率。aug_prob默认是0.4.由于tia数据增强特殊性，默认不采用，可以通过添加use_tia设置，使tia数据增强生效。详细设置可以参考[ISSUE 1744](https://github.com/PaddlePaddle/PaddleOCR/issues/1744)。
#### Q3.4.31: 采用Paddle-Lite进行端侧部署，出现问题，环境没问题。
#### Q3.4.28: PP-OCR系统中，文本检测的结果有置信度吗？
**A**：如果你的预测库是自己编译的，那么你的nb文件也要自己编译，用同一个lite版本。不能直接用下载的nb文件，因为版本不同。
**A**：文本检测的结果有置信度，由于推理过程中没有使用，所以没有显示的返回到最终结果中。如果需要文本检测结果的置信度，可以在[文本检测DB的后处理代码](../../ppocr/postprocess/db_postprocess.py)的155行，添加scores信息。这样，在[检测预测代码](../../tools/infer/predict_det.py)的197行，就可以拿到文本检测的scores信息。
#### Q3.4.32: PaddleOCR的模型支持onnx转换吗？
#### Q3.4.29: DB文本检测，特征提取网络金字塔构建的部分代码在哪儿？
**A**：特征提取网络金字塔构建的部分:[代码位置](../../ppocr/modeling/necks/db_fpn.py)。ppocr/modeling文件夹里面是组网相关的代码，其中architectures是文本检测或者文本识别整体流程代码；backbones是骨干网络相关代码；necks是类似与FPN的颈函数代码；heads是提取文本检测或者文本识别预测结果相关的头函数；transforms是类似于TPS特征预处理模块。更多的信息可以参考[代码组织结构](./tree.md)。
**A**：我们目前已经通过Paddle2ONNX来支持各模型套件的转换，PaddleOCR基于PaddlePaddle 2.0的版本（dygraph分支）已经支持导出为ONNX，欢迎关注Paddle2ONNX，了解更多项目的进展：
Paddle2ONNX项目：https://github.com/PaddlePaddle/Paddle2ONNX
Paddle2ONNX支持转换的[模型列表](https://github.com/PaddlePaddle/Paddle2ONNX/blob/develop/docs/zh/model_zoo.md#%E5%9B%BE%E5%83%8Focr)
<a name="OCR精选10个问题"></a>
## 【精选】OCR精选10个问题
......@@ -474,9 +475,18 @@ StyleText的用途主要是：提取style_image中的字体、背景等style信
例如识别身份证照片，可以先匹配"姓名"，"性别"等关键字，根据这些关键字的坐标去推测其他信息的位置，再与识别的结果匹配。
#### Q3.1.36 如何识别竹简上的古文？
**A**：对于字符都是普通的汉字字符的情况，只要标注足够的数据，finetune模型就可以了。如果数据量不足，您可以尝试StyleText工具。
而如果使用的字符是特殊的古文字、甲骨文、象形文字等，那么首先需要构建一个古文字的字典，之后再进行训练。
#### Q3.1.37: 小语种模型只有识别模型，没有检测模型吗？
**A**：小语种（包括纯英文数字）的检测模型和中文的检测模型是共用的，在训练中文检测模型时加入了多语言数据。https://github.com/PaddlePaddle/PaddleOCR/blob/dygraph/doc/doc_en/models_list_en.md#1-text-detection-model。
#### Q3.1.38: module 'paddle.distributed' has no attribute ‘get_rank’。
**A**：Paddle版本问题，请安装2.0版本Paddle：pip install paddlepaddle==2.0.0rc1。
<a name="数据集3"></a>
### 数据集
......@@ -854,3 +864,17 @@ img = cv.imdecode(img_array, -1)
#### Q3.4.29: DB文本检测，特征提取网络金字塔构建的部分代码在哪儿？
**A**：特征提取网络金字塔构建的部分:[代码位置](../../ppocr/modeling/necks/db_fpn.py)。ppocr/modeling文件夹里面是组网相关的代码，其中architectures是文本检测或者文本识别整体流程代码；backbones是骨干网络相关代码；necks是类似与FPN的颈函数代码；heads是提取文本检测或者文本识别预测结果相关的头函数；transforms是类似于TPS特征预处理模块。更多的信息可以参考[代码组织结构](./tree.md)。
#### Q3.4.30: PaddleOCR是否支持在华为鲲鹏920CPU上部署？
**A**：目前Paddle的预测库是支持华为鲲鹏920CPU的，但是OCR还没在这些芯片上测试过，可以自己调试，有问题反馈给我们。
#### Q3.4.31: 采用Paddle-Lite进行端侧部署，出现问题，环境没问题。
**A**：如果你的预测库是自己编译的，那么你的nb文件也要自己编译，用同一个lite版本。不能直接用下载的nb文件，因为版本不同。
#### Q3.4.32: PaddleOCR的模型支持onnx转换吗？
**A**：我们目前已经通过Paddle2ONNX来支持各模型套件的转换，PaddleOCR基于PaddlePaddle 2.0的版本（dygraph分支）已经支持导出为ONNX，欢迎关注Paddle2ONNX，了解更多项目的进展：
Paddle2ONNX项目：https://github.com/PaddlePaddle/Paddle2ONNX
Paddle2ONNX支持转换的[模型列表](https://github.com/PaddlePaddle/Paddle2ONNX/blob/develop/docs/zh/model_zoo.md#%E5%9B%BE%E5%83%8Focr)
\ No newline at end of file
doc/doc_ch/algorithm_overview.md
浏览文件 @ 5e9fb50d
......@@ -14,11 +14,10 @@ PaddleOCR开源的文本检测算法列表：
- [x] SAST([paper](https://arxiv.org/abs/1908.05498))[4]
在ICDAR2015文本检测公开数据集上，算法效果如下：
|模型|骨干网络|precision|recall|Hmean|下载链接|
| --- | --- | --- | --- | --- | --- |
|EAST|ResNet50_vd|88.76%|81.36%|84.90%|[下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/det_r50_vd_east_v2.0_train.tar)|
|EAST|MobileNetV3|78.24%|79.15%|78.69%|[下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/det_mv3_east_v2.0_train.tar)|
|EAST|ResNet50_vd|85.80%|86.71%|86.25%|[下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/det_r50_vd_east_v2.0_train.tar)|
|EAST|MobileNetV3|79.42%|80.64%|80.03%|[下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/det_mv3_east_v2.0_train.tar)|
|DB|ResNet50_vd|86.41%|78.72%|82.38%|[下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/det_r50_vd_db_v2.0_train.tar)|
|DB|MobileNetV3|77.29%|73.08%|75.12%|[下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/det_mv3_db_v2.0_train.tar)|
|SAST|ResNet50_vd|91.39%|83.77%|87.42%|[下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/det_r50_vd_sast_icdar15_v2.0_train.tar)|
......@@ -40,17 +39,19 @@ PaddleOCR文本检测算法的训练和使用请参考文档教程中[模型训
PaddleOCR基于动态图开源的文本识别算法列表：
- [x] CRNN([paper](https://arxiv.org/abs/1507.05717))[7]（ppocr推荐）
- [x] Rosetta([paper](https://arxiv.org/abs/1910.05085))[10]
- [ ] STAR-Net([paper](http://www.bmva.org/bmvc/2016/papers/paper043/index.html))[11] coming soon
- [x] STAR-Net([paper](http://www.bmva.org/bmvc/2016/papers/paper043/index.html))[11]
- [ ] RARE([paper](https://arxiv.org/abs/1603.03915v1))[12] coming soon
- [ ] SRN([paper](https://arxiv.org/abs/2003.12294))[5] coming soon
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：
|模型|骨干网络|Avg Accuracy|模型存储命名|下载链接|
|-|-|-|-|-|
|---|---|---|---|---|
|Rosetta|Resnet34_vd|80.9%|rec_r34_vd_none_none_ctc|[下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_r34_vd_none_none_ctc_v2.0_train.tar)|
|Rosetta|MobileNetV3|78.05%|rec_mv3_none_none_ctc|[下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_mv3_none_none_ctc_v2.0_train.tar)|
|CRNN|Resnet34_vd|82.76%|rec_r34_vd_none_bilstm_ctc|[下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_r34_vd_none_bilstm_ctc_v2.0_train.tar)|
|CRNN|MobileNetV3|79.97%|rec_mv3_none_bilstm_ctc|[下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_mv3_none_bilstm_ctc_v2.0_train.tar)|
|StarNet|Resnet34_vd|84.44%|rec_r34_vd_tps_bilstm_ctc|[下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_r34_vd_tps_bilstm_ctc_v2.0_train.tar)|
|StarNet|MobileNetV3|81.42%|rec_mv3_tps_bilstm_ctc|[下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_mv3_tps_bilstm_ctc_v2.0_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/inference.md
浏览文件 @ 5e9fb50d
......@@ -352,10 +352,10 @@ Predicts of ./doc/imgs_words/ch/word_4.jpg:['0', 0.9999982]
```
# 使用方向分类器
python3 tools/infer/predict_system.py --image_dir="./doc/imgs/2.jpg" --det_model_dir="./inference/det_db/" --cls_model_dir="./inference/cls/" --rec_model_dir="./inference/rec_crnn/" --use_angle_cls=true
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/2.jpg" --det_model_dir="./inference/det_db/" --rec_model_dir="./inference/rec_crnn/" --use_angle_cls=false
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
```
......@@ -364,7 +364,7 @@ python3 tools/infer/predict_system.py --image_dir="./doc/imgs/2.jpg" --det_model
执行命令后，识别结果图像如下：
![](../imgs_results/2.jpg)
![](../imgs_results/system_res_00018069.jpg)
<a name="其他模型推理"></a>
### 2. 其他模型推理
......@@ -381,4 +381,4 @@ python3 tools/infer/predict_system.py --image_dir="./doc/imgs_en/img_10.jpg" --d
执行命令后，识别结果图像如下：
(coming soon)
![](../imgs_results/img_10_east_starnet.jpg)
doc/doc_ch/models_list.md
浏览文件 @ 5e9fb50d
## OCR模型列表（V2.0，2020年12月12日更新）
## OCR模型列表（V2.0，2021年1月20日更新）
**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。
- [一、文本检测模型](#文本检测模型)
......@@ -22,7 +22,7 @@ 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_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)| |推理模型 (coming soon) / slim模型 (coming soon)|
|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)| |推理模型 (coming soon) / 训练模型 (coming soon)|
|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)|
......@@ -35,7 +35,7 @@ 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_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)| |推理模型 (coming soon) / slim模型 (coming soon) |
|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)| |[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)|3.71M|[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) |
......@@ -46,7 +46,7 @@ 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|
| --- | --- | --- | --- | --- |
|en_number_mobile_slim_v2.0_rec|slim裁剪量化版超轻量模型，支持英文、数字识别|[rec_en_number_lite_train.yml](../../configs/rec/multi_language/rec_en_number_lite_train.yml)| | 推理模型 (coming soon) / slim模型 (coming soon) |
|en_number_mobile_slim_v2.0_rec|slim裁剪量化版超轻量模型，支持英文、数字识别|[rec_en_number_lite_train.yml](../../configs/rec/multi_language/rec_en_number_lite_train.yml)| | [推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/en_number_mobile_v2.0_rec_slim_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/en_number_mobile_v2.0_rec_slim_train.tar) |
|en_number_mobile_v2.0_rec|原始超轻量模型，支持英文、数字识别|[rec_en_number_lite_train.yml](../../configs/rec/multi_language/rec_en_number_lite_train.yml)|2.56M|[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/en_number_mobile_v2.0_rec_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/en_number_mobile_v2.0_rec_train.tar) |
<a name="多语言识别模型"></a>
......@@ -55,7 +55,7 @@ PaddleOCR提供的可下载模型包括`推理模型`、`训练模型`、`预训
**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 PaddleOCR/configs/rec/multi_language/
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：
......@@ -64,7 +64,7 @@ python3 generate_multi_language_configs.py --help
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 PaddleOCR/configs/rec/multi_language/
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
```
......@@ -77,6 +77,8 @@ python3 generate_multi_language_configs.py -l it
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
......@@ -112,7 +114,7 @@ python3 generate_multi_language_configs.py -l it \
| uk_mobile_v2.0_rec |乌克兰文识别|rec_uk_lite_train.yml|2.63M|[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/uk_mobile_v2.0_rec_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/uk_mobile_v2.0_rec_train.tar) |
| be_mobile_v2.0_rec |白俄罗斯文识别|rec_be_lite_train.yml|2.63M|[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/be_mobile_v2.0_rec_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/be_mobile_v2.0_rec_train.tar) |
| te_mobile_v2.0_rec |泰卢固文识别|rec_te_lite_train.yml|2.63M|[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/te_mobile_v2.0_rec_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/te_mobile_v2.0_rec_train.tar) |
| ka_mobile_v2.0_rec |卡纳达文识别|[rec_ka_lite_train.yml|2.63M|[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ka_mobile_v2.0_rec_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ka_mobile_v2.0_rec_train.tar) |
| ka_mobile_v2.0_rec |卡纳达文识别|rec_ka_lite_train.yml|2.63M|[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ka_mobile_v2.0_rec_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ka_mobile_v2.0_rec_train.tar) |
| ta_mobile_v2.0_rec |泰米尔文识别|rec_ta_lite_train.yml|2.63M|[推理模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ta_mobile_v2.0_rec_infer.tar) / [训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ta_mobile_v2.0_rec_train.tar) |
......@@ -121,5 +123,5 @@ 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)| |推理模型 (coming soon) / 训练模型 / slim模型 |
|ch_ppocr_mobile_slim_v2.0_cls|slim量化版模型|[cls_mv3.yml](../../configs/cls/cls_mv3.yml)| |[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) |
doc/doc_en/algorithm_overview_en.md
浏览文件 @ 5e9fb50d
......@@ -19,8 +19,8 @@ On the ICDAR2015 dataset, the text detection result is as follows:
|Model|Backbone|precision|recall|Hmean|Download link|
| --- | --- | --- | --- | --- | --- |
|EAST|ResNet50_vd|88.76%|81.36%|84.90%|[Download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/det_r50_vd_east_v2.0_train.tar)|
|EAST|MobileNetV3|78.24%|79.15%|78.69%|[Download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/det_mv3_east_v2.0_train.tar)|
|EAST|ResNet50_vd|85.80%|86.71%|86.25%|[Download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/det_r50_vd_east_v2.0_train.tar)|
|EAST|MobileNetV3|79.42%|80.64%|80.03%|[Download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/det_mv3_east_v2.0_train.tar)|
|DB|ResNet50_vd|86.41%|78.72%|82.38%|[Download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/det_r50_vd_db_v2.0_train.tar)|
|DB|MobileNetV3|77.29%|73.08%|75.12%|[Download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/det_mv3_db_v2.0_train.tar)|
|SAST|ResNet50_vd|91.39%|83.77%|87.42%|[Download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/det_r50_vd_sast_icdar15_v2.0_train.tar)|
......@@ -41,17 +41,19 @@ For the training guide and use of PaddleOCR text detection algorithms, please re
PaddleOCR open-source text recognition algorithms list:
- [x] CRNN([paper](https://arxiv.org/abs/1507.05717))[7]
- [x] Rosetta([paper](https://arxiv.org/abs/1910.05085))[10]
- [ ] STAR-Net([paper](http://www.bmva.org/bmvc/2016/papers/paper043/index.html))[11] coming soon
- [x] STAR-Net([paper](http://www.bmva.org/bmvc/2016/papers/paper043/index.html))[11]
- [ ] RARE([paper](https://arxiv.org/abs/1603.03915v1))[12] coming soon
- [ ] SRN([paper](https://arxiv.org/abs/2003.12294))[5] coming soon
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:
|Model|Backbone|Avg Accuracy|Module combination|Download link|
|-|-|-|-|-|
|---|---|---|---|---|
|Rosetta|Resnet34_vd|80.9%|rec_r34_vd_none_none_ctc|[Download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_r34_vd_none_none_ctc_v2.0_train.tar)|
|Rosetta|MobileNetV3|78.05%|rec_mv3_none_none_ctc|[Download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_mv3_none_none_ctc_v2.0_train.tar)|
|CRNN|Resnet34_vd|82.76%|rec_r34_vd_none_bilstm_ctc|[Download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_r34_vd_none_bilstm_ctc_v2.0_train.tar)|
|CRNN|MobileNetV3|79.97%|rec_mv3_none_bilstm_ctc|[Download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_mv3_none_bilstm_ctc_v2.0_train.tar)|
|StarNet|Resnet34_vd|84.44%|rec_r34_vd_tps_bilstm_ctc|[Download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_r34_vd_tps_bilstm_ctc_v2.0_train.tar)|
|StarNet|MobileNetV3|81.42%|rec_mv3_tps_bilstm_ctc|[Download link](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_mv3_tps_bilstm_ctc_v2.0_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/inference_en.md
浏览文件 @ 5e9fb50d
......@@ -366,15 +366,15 @@ When performing prediction, you need to specify the path of a single image or a
```
# use direction classifier
python3 tools/infer/predict_system.py --image_dir="./doc/imgs/2.jpg" --det_model_dir="./inference/det_db/" --cls_model_dir="./inference/cls/" --rec_model_dir="./inference/rec_crnn/" --use_angle_cls=true
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/2.jpg" --det_model_dir="./inference/det_db/" --rec_model_dir="./inference/rec_crnn/"
python3 tools/infer/predict_system.py --image_dir="./doc/imgs/00018069.jpg" --det_model_dir="./inference/det_db/" --rec_model_dir="./inference/rec_crnn/"
```
After executing the command, the recognition result image is as follows:
![](../imgs_results/2.jpg)
![](../imgs_results/system_res_00018069.jpg)
<a name="OTHER_MODELS"></a>
### 2. OTHER MODELS
......@@ -391,4 +391,4 @@ python3 tools/infer/predict_system.py --image_dir="./doc/imgs_en/img_10.jpg" --d
After executing the command, the recognition result image is as follows:
(coming soon)
![](../imgs_results/img_10_east_starnet.jpg)
doc/doc_en/models_list_en.md
浏览文件 @ 5e9fb50d
......@@ -33,7 +33,7 @@ The downloadable models provided by PaddleOCR include `inference model`, `traine
|model name|description|config|model size|download|
| --- | --- | --- | --- | --- |
|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)| |inference model (coming soon) / slim model (coming soon) |
|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)| | [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)|3.71M|[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) |
......@@ -45,7 +45,7 @@ The downloadable models provided by PaddleOCR include `inference model`, `traine
|model name|description|config|model size|download|
| --- | --- | --- | --- | --- |
|en_number_mobile_slim_v2.0_rec|Slim pruned and quantized lightweight model, supporting English and number recognition|[rec_en_number_lite_train.yml](../../configs/rec/multi_language/rec_en_number_lite_train.yml)| |inference model (coming soon ) / slim model (coming soon) |
|en_number_mobile_slim_v2.0_rec|Slim pruned and quantized lightweight model, supporting English and number recognition|[rec_en_number_lite_train.yml](../../configs/rec/multi_language/rec_en_number_lite_train.yml)| | [inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/en_number_mobile_v2.0_rec_slim_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/en_number_mobile_v2.0_rec_slim_train.tar) |
|en_number_mobile_v2.0_rec|Original lightweight model, supporting English and number recognition|[rec_en_number_lite_train.yml](../../configs/rec/multi_language/rec_en_number_lite_train.yml)|2.56M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/en_number_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/en_number_mobile_v2.0_rec_train.tar) |
<a name="Multilingual"></a>
......@@ -54,6 +54,8 @@ The downloadable models provided by PaddleOCR include `inference model`, `traine
**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
```
......@@ -62,7 +64,7 @@ Take the Italian configuration file as an example：
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 PaddleOCR/configs/rec/multi_language/
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
......@@ -74,6 +76,8 @@ If you want to train your own model, you can prepare the training set file, veri
- 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
......@@ -92,27 +96,27 @@ python3 generate_multi_language_configs.py -l it \
| german_mobile_v2.0_rec |Lightweight model for French recognition|[rec_german_lite_train.yml](../../configs/rec/multi_language/rec_german_lite_train.yml)|2.65M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/german_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/german_mobile_v2.0_rec_train.tar) |
| korean_mobile_v2.0_rec |Lightweight model for Korean recognition|[rec_korean_lite_train.yml](../../configs/rec/multi_language/rec_korean_lite_train.yml)|3.9M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/korean_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/korean_mobile_v2.0_rec_train.tar) |
| japan_mobile_v2.0_rec |Lightweight model for Japanese recognition|[rec_japan_lite_train.yml](../../configs/rec/multi_language/rec_japan_lite_train.yml)|4.23M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/japan_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/japan_mobile_v2.0_rec_train.tar) |
| it_mobile_v2.0_rec |Lightweight model for Italian recognition|rec_it_lite_train.yml|2.53M|[inference mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/it_mobile_v2.0_rec_infer.tar) / [trained mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/it_mobile_v2.0_rec_train.tar) |
| xi_mobile_v2.0_rec |Lightweight model for Spanish recognition|rec_xi_lite_train.yml|2.53M|[inference mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/xi_mobile_v2.0_rec_infer.tar) / [trained mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/xi_mobile_v2.0_rec_train.tar) |
| pu_mobile_v2.0_rec |Lightweight model for Portuguese recognition|rec_pu_lite_train.yml|2.63M|[inference mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/pu_mobile_v2.0_rec_infer.tar) / [trained mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/pu_mobile_v2.0_rec_train.tar) |
| ru_mobile_v2.0_rec |Lightweight model for Russia recognition|rec_ru_lite_train.yml|2.63M|[inference mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ru_mobile_v2.0_rec_infer.tar) / [trained mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ru_mobile_v2.0_rec_train.tar) |
| ar_mobile_v2.0_rec |Lightweight model for Arabic recognition|rec_ar_lite_train.yml|2.53M|[inference mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ar_mobile_v2.0_rec_infer.tar) / [trained mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ar_mobile_v2.0_rec_train.tar) |
| hi_mobile_v2.0_rec |Lightweight model for Hindi recognition|rec_hi_lite_train.yml|2.63M|[inference mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/hi_mobile_v2.0_rec_infer.tar) / [trained mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/hi_mobile_v2.0_rec_train.tar) |
| chinese_cht_mobile_v2.0_rec |Lightweight model for chinese traditional recognition|rec_chinese_cht_lite_train.yml|5.63M|[inference mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/chinese_cht_mobile_v2.0_rec_infer.tar) / [trained mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/chinese_cht_mobile_v2.0_rec_train.tar) |
| ug_mobile_v2.0_rec |Lightweight model for Uyghur recognition|rec_ug_lite_train.yml|2.63M|[inference mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ug_mobile_v2.0_rec_infer.tar) / [trained mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ug_mobile_v2.0_rec_train.tar) |
| fa_mobile_v2.0_rec |Lightweight model for Persian recognition|rec_fa_lite_train.yml|2.63M|[inference mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/fa_mobile_v2.0_rec_infer.tar) / [trained mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/fa_mobile_v2.0_rec_train.tar) |
| ur_mobile_v2.0_rec |Lightweight model for Urdu recognition|rec_ur_lite_train.yml|2.63M|[inference mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ur_mobile_v2.0_rec_infer.tar) / [trained mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ur_mobile_v2.0_rec_train.tar) |
| rs_mobile_v2.0_rec |Lightweight model for Serbian(latin) recognition|rec_rs_lite_train.yml|2.53M|[inference mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/rs_mobile_v2.0_rec_infer.tar) / [trained mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/rs_mobile_v2.0_rec_train.tar) |
| oc_mobile_v2.0_rec |Lightweight model for Occitan recognition|rec_oc_lite_train.yml|2.53M|[inference mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/oc_mobile_v2.0_rec_infer.tar) / [trained mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/oc_mobile_v2.0_rec_train.tar) |
| mr_mobile_v2.0_rec |Lightweight model for Marathi recognition|rec_mr_lite_train.yml|2.63M|[inference mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/mr_mobile_v2.0_rec_infer.tar) / [trained mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/mr_mobile_v2.0_rec_train.tar) |
| ne_mobile_v2.0_rec |Lightweight model for Nepali recognition|rec_ne_lite_train.yml|2.63M|[inference mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ne_mobile_v2.0_rec_infer.tar) / [trained mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ne_mobile_v2.0_rec_train.tar) |
| rsc_mobile_v2.0_rec |Lightweight model for Serbian(cyrillic) recognition|rec_rsc_lite_train.yml|2.63M|[inference mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/rsc_mobile_v2.0_rec_infer.tar) / [trained mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/rsc_mobile_v2.0_rec_train.tar) |
| bg_mobile_v2.0_rec |Lightweight model for Bulgarian recognition|rec_bg_lite_train.yml|2.63M|[inference mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/bg_mobile_v2.0_rec_infer.tar) / [trained mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/bg_mobile_v2.0_rec_train.tar) |
| uk_mobile_v2.0_rec |Lightweight model for Ukranian recognition|rec_uk_lite_train.yml|2.63M|[inference mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/uk_mobile_v2.0_rec_infer.tar) / [trained mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/uk_mobile_v2.0_rec_train.tar) |
| be_mobile_v2.0_rec |Lightweight model for Belarusian recognition|rec_be_lite_train.yml|2.63M|[inference mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/be_mobile_v2.0_rec_infer.tar) / [trained mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/be_mobile_v2.0_rec_train.tar) |
| te_mobile_v2.0_rec |Lightweight model for Telugu recognition|rec_te_lite_train.yml|2.63M|[inference mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/te_mobile_v2.0_rec_infer.tar) / [trained mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/te_mobile_v2.0_rec_train.tar) |
| ka_mobile_v2.0_rec |Lightweight model for Kannada recognition|[rec_ka_lite_train.yml|2.63M|[inference mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ka_mobile_v2.0_rec_infer.tar) / [trained mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ka_mobile_v2.0_rec_train.tar) |
| ta_mobile_v2.0_rec |Lightweight model for Tamil recognition|rec_ta_lite_train.yml|2.63M|[inference mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ta_mobile_v2.0_rec_infer.tar) / [trained mode](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ta_mobile_v2.0_rec_train.tar) |
| it_mobile_v2.0_rec |Lightweight model for Italian recognition|rec_it_lite_train.yml|2.53M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/it_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/it_mobile_v2.0_rec_train.tar) |
| xi_mobile_v2.0_rec |Lightweight model for Spanish recognition|rec_xi_lite_train.yml|2.53M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/xi_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/xi_mobile_v2.0_rec_train.tar) |
| pu_mobile_v2.0_rec |Lightweight model for Portuguese recognition|rec_pu_lite_train.yml|2.63M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/pu_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/pu_mobile_v2.0_rec_train.tar) |
| ru_mobile_v2.0_rec |Lightweight model for Russia recognition|rec_ru_lite_train.yml|2.63M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ru_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ru_mobile_v2.0_rec_train.tar) |
| ar_mobile_v2.0_rec |Lightweight model for Arabic recognition|rec_ar_lite_train.yml|2.53M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ar_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ar_mobile_v2.0_rec_train.tar) |
| hi_mobile_v2.0_rec |Lightweight model for Hindi recognition|rec_hi_lite_train.yml|2.63M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/hi_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/hi_mobile_v2.0_rec_train.tar) |
| chinese_cht_mobile_v2.0_rec |Lightweight model for chinese traditional recognition|rec_chinese_cht_lite_train.yml|5.63M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/chinese_cht_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/chinese_cht_mobile_v2.0_rec_train.tar) |
| ug_mobile_v2.0_rec |Lightweight model for Uyghur recognition|rec_ug_lite_train.yml|2.63M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ug_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ug_mobile_v2.0_rec_train.tar) |
| fa_mobile_v2.0_rec |Lightweight model for Persian recognition|rec_fa_lite_train.yml|2.63M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/fa_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/fa_mobile_v2.0_rec_train.tar) |
| ur_mobile_v2.0_rec |Lightweight model for Urdu recognition|rec_ur_lite_train.yml|2.63M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ur_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ur_mobile_v2.0_rec_train.tar) |
| rs_mobile_v2.0_rec |Lightweight model for Serbian(latin) recognition|rec_rs_lite_train.yml|2.53M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/rs_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/rs_mobile_v2.0_rec_train.tar) |
| oc_mobile_v2.0_rec |Lightweight model for Occitan recognition|rec_oc_lite_train.yml|2.53M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/oc_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/oc_mobile_v2.0_rec_train.tar) |
| mr_mobile_v2.0_rec |Lightweight model for Marathi recognition|rec_mr_lite_train.yml|2.63M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/mr_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/mr_mobile_v2.0_rec_train.tar) |
| ne_mobile_v2.0_rec |Lightweight model for Nepali recognition|rec_ne_lite_train.yml|2.63M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ne_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ne_mobile_v2.0_rec_train.tar) |
| rsc_mobile_v2.0_rec |Lightweight model for Serbian(cyrillic) recognition|rec_rsc_lite_train.yml|2.63M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/rsc_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/rsc_mobile_v2.0_rec_train.tar) |
| bg_mobile_v2.0_rec |Lightweight model for Bulgarian recognition|rec_bg_lite_train.yml|2.63M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/bg_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/bg_mobile_v2.0_rec_train.tar) |
| uk_mobile_v2.0_rec |Lightweight model for Ukranian recognition|rec_uk_lite_train.yml|2.63M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/uk_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/uk_mobile_v2.0_rec_train.tar) |
| be_mobile_v2.0_rec |Lightweight model for Belarusian recognition|rec_be_lite_train.yml|2.63M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/be_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/be_mobile_v2.0_rec_train.tar) |
| te_mobile_v2.0_rec |Lightweight model for Telugu recognition|rec_te_lite_train.yml|2.63M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/te_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/te_mobile_v2.0_rec_train.tar) |
| ka_mobile_v2.0_rec |Lightweight model for Kannada recognition|rec_ka_lite_train.yml|2.63M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ka_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ka_mobile_v2.0_rec_train.tar) |
| ta_mobile_v2.0_rec |Lightweight model for Tamil recognition|rec_ta_lite_train.yml|2.63M|[inference model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ta_mobile_v2.0_rec_infer.tar) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/ta_mobile_v2.0_rec_train.tar) |
<a name="Angle"></a>
......@@ -120,6 +124,5 @@ python3 generate_multi_language_configs.py -l it \
|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)| |inference model (coming soon) / trained model / slim model|
|ch_ppocr_mobile_slim_v2.0_cls|Slim quantized model|[cls_mv3.yml](../../configs/cls/cls_mv3.yml)| | [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) |
doc/fonts/arabic.ttf 0 → 100644
浏览文件 @ 5e9fb50d
文件已添加
doc/fonts/hindi.ttf 0 → 100644
浏览文件 @ 5e9fb50d
文件已添加
doc/fonts/latin.ttf 0 → 100644
浏览文件 @ 5e9fb50d
文件已添加
doc/fonts/nepali.ttf 0 → 100644
浏览文件 @ 5e9fb50d
文件已添加
doc/fonts/tamil.ttf 0 → 100644
浏览文件 @ 5e9fb50d
文件已添加
doc/fonts/telugu.ttf 0 → 100644
浏览文件 @ 5e9fb50d
文件已添加
doc/fonts/urdu.ttf 0 → 100644
浏览文件 @ 5e9fb50d
文件已添加
doc/fonts/uyghur.ttf 0 → 100644
浏览文件 @ 5e9fb50d
文件已添加
doc/joinus.PNG
查看替换文件 @ 45117f90 浏览文件 @ 5e9fb50d
doc/joinus.PNG

108.9 KB | W: | H:

doc/joinus.PNG

107.3 KB | W: | H:

doc/joinus.PNG
doc/joinus.PNG
doc/joinus.PNG
doc/joinus.PNG
  • 2-up
  • Swipe
  • Onion skin
paddleocr.py
浏览文件 @ 5e9fb50d
......@@ -290,7 +290,9 @@ class PaddleOCR(predict_system.TextSystem):
image_file = img
img, flag = check_and_read_gif(image_file)
if not flag:
img = cv2.imread(image_file)
with open(image_file, 'rb') as f:
np_arr = np.frombuffer(f.read(), dtype=np.uint8)
img = cv2.imdecode(np_arr, cv2.IMREAD_COLOR)
if img is None:
logger.error("error in loading image:{}".format(image_file))
return None
......
ppocr/data/__init__.py
浏览文件 @ 5e9fb50d
......@@ -51,7 +51,7 @@ signal.signal(signal.SIGINT, term_mp)
signal.signal(signal.SIGTERM, term_mp)
def build_dataloader(config, mode, device, logger):
def build_dataloader(config, mode, device, logger, seed=None):
config = copy.deepcopy(config)
support_dict = ['SimpleDataSet', 'LMDBDateSet']
......@@ -61,7 +61,7 @@ def build_dataloader(config, mode, device, logger):
assert mode in ['Train', 'Eval', 'Test'
], "Mode should be Train, Eval or Test."
dataset = eval(module_name)(config, mode, logger)
dataset = eval(module_name)(config, mode, logger, seed)
loader_config = config[mode]['loader']
batch_size = loader_config['batch_size_per_card']
drop_last = loader_config['drop_last']
......
ppocr/data/imaug/sast_process.py
浏览文件 @ 5e9fb50d
......@@ -24,11 +24,11 @@ __all__ = ['SASTProcessTrain']
class SASTProcessTrain(object):
def __init__(self,
image_shape = [512, 512],
min_crop_size = 24,
min_crop_side_ratio = 0.3,
min_text_size = 10,
max_text_size = 512,
image_shape=[512, 512],
min_crop_size=24,
min_crop_side_ratio=0.3,
min_text_size=10,
max_text_size=512,
**kwargs):
self.input_size = image_shape[1]
self.min_crop_size = min_crop_size
......@@ -42,12 +42,10 @@ class SASTProcessTrain(object):
:param poly:
:return:
"""
edge = [
(poly[1][0] - poly[0][0]) * (poly[1][1] + poly[0][1]),
edge = [(poly[1][0] - poly[0][0]) * (poly[1][1] + poly[0][1]),
(poly[2][0] - poly[1][0]) * (poly[2][1] + poly[1][1]),
(poly[3][0] - poly[2][0]) * (poly[3][1] + poly[2][1]),
(poly[0][0] - poly[3][0]) * (poly[0][1] + poly[3][1])
]
(poly[0][0] - poly[3][0]) * (poly[0][1] + poly[3][1])]
return np.sum(edge) / 2.
def gen_quad_from_poly(self, poly):
......@@ -57,7 +55,8 @@ class SASTProcessTrain(object):
point_num = poly.shape[0]
min_area_quad = np.zeros((4, 2), dtype=np.float32)
if True:
rect = cv2.minAreaRect(poly.astype(np.int32)) # (center (x,y), (width, height), angle of rotation)
rect = cv2.minAreaRect(poly.astype(
np.int32)) # (center (x,y), (width, height), angle of rotation)
center_point = rect[0]
box = np.array(cv2.boxPoints(rect))
......@@ -103,11 +102,14 @@ class SASTProcessTrain(object):
if tag == False:
print('poly in wrong direction')
tag = True # reversed cases should be ignore
poly = poly[(0, 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1), :]
poly = poly[(0, 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2,
1), :]
quad = quad[(0, 3, 2, 1), :]
len_w = np.linalg.norm(quad[0] - quad[1]) + np.linalg.norm(quad[3] - quad[2])
len_h = np.linalg.norm(quad[0] - quad[3]) + np.linalg.norm(quad[1] - quad[2])
len_w = np.linalg.norm(quad[0] - quad[1]) + np.linalg.norm(quad[3] -
quad[2])
len_h = np.linalg.norm(quad[0] - quad[3]) + np.linalg.norm(quad[1] -
quad[2])
hv_tag = 1
if len_w * 2.0 < len_h:
......@@ -116,9 +118,16 @@ class SASTProcessTrain(object):
validated_polys.append(poly)
validated_tags.append(tag)
hv_tags.append(hv_tag)
return np.array(validated_polys), np.array(validated_tags), np.array(hv_tags)
return np.array(validated_polys), np.array(validated_tags), np.array(
hv_tags)
def crop_area(self, im, polys, tags, hv_tags, crop_background=False, max_tries=25):
def crop_area(self,
im,
polys,
tags,
hv_tags,
crop_background=False,
max_tries=25):
"""
make random crop from the input image
:param im:
......@@ -137,10 +146,10 @@ class SASTProcessTrain(object):
poly = np.round(poly, decimals=0).astype(np.int32)
minx = np.min(poly[:, 0])
maxx = np.max(poly[:, 0])
w_array[minx + pad_w: maxx + pad_w] = 1
w_array[minx + pad_w:maxx + pad_w] = 1
miny = np.min(poly[:, 1])
maxy = np.max(poly[:, 1])
h_array[miny + pad_h: maxy + pad_h] = 1
h_array[miny + pad_h:maxy + pad_h] = 1
# ensure the cropped area not across a text
h_axis = np.where(h_array == 0)[0]
w_axis = np.where(w_array == 0)[0]
......@@ -166,17 +175,18 @@ class SASTProcessTrain(object):
if polys.shape[0] != 0:
poly_axis_in_area = (polys[:, :, 0] >= xmin) & (polys[:, :, 0] <= xmax) \
& (polys[:, :, 1] >= ymin) & (polys[:, :, 1] <= ymax)
selected_polys = np.where(np.sum(poly_axis_in_area, axis=1) == 4)[0]
selected_polys = np.where(
np.sum(poly_axis_in_area, axis=1) == 4)[0]
else:
selected_polys = []
if len(selected_polys) == 0:
# no text in this area
if crop_background:
return im[ymin : ymax + 1, xmin : xmax + 1, :], \
polys[selected_polys], tags[selected_polys], hv_tags[selected_polys], txts
polys[selected_polys], tags[selected_polys], hv_tags[selected_polys]
else:
continue
im = im[ymin: ymax + 1, xmin: xmax + 1, :]
im = im[ymin:ymax + 1, xmin:xmax + 1, :]
polys = polys[selected_polys]
tags = tags[selected_polys]
hv_tags = hv_tags[selected_polys]
......@@ -192,18 +202,28 @@ class SASTProcessTrain(object):
width_list = []
height_list = []
for quad in poly_quads:
quad_w = (np.linalg.norm(quad[0] - quad[1]) + np.linalg.norm(quad[2] - quad[3])) / 2.0
quad_h = (np.linalg.norm(quad[0] - quad[3]) + np.linalg.norm(quad[2] - quad[1])) / 2.0
quad_w = (np.linalg.norm(quad[0] - quad[1]) +
np.linalg.norm(quad[2] - quad[3])) / 2.0
quad_h = (np.linalg.norm(quad[0] - quad[3]) +
np.linalg.norm(quad[2] - quad[1])) / 2.0
width_list.append(quad_w)
height_list.append(quad_h)
norm_width = max(sum(width_list) / (len(width_list) + 1e-6), 1.0)
average_height = max(sum(height_list) / (len(height_list) + 1e-6), 1.0)
for quad in poly_quads:
direct_vector_full = ((quad[1] + quad[2]) - (quad[0] + quad[3])) / 2.0
direct_vector = direct_vector_full / (np.linalg.norm(direct_vector_full) + 1e-6) * norm_width
direction_label = tuple(map(float, [direct_vector[0], direct_vector[1], 1.0 / (average_height + 1e-6)]))
cv2.fillPoly(direction_map, quad.round().astype(np.int32)[np.newaxis, :, :], direction_label)
direct_vector_full = (
(quad[1] + quad[2]) - (quad[0] + quad[3])) / 2.0
direct_vector = direct_vector_full / (
np.linalg.norm(direct_vector_full) + 1e-6) * norm_width
direction_label = tuple(
map(float, [
direct_vector[0], direct_vector[1], 1.0 / (average_height +
1e-6)
]))
cv2.fillPoly(direction_map,
quad.round().astype(np.int32)[np.newaxis, :, :],
direction_label)
return direction_map
def calculate_average_height(self, poly_quads):
......@@ -211,13 +231,19 @@ class SASTProcessTrain(object):
"""
height_list = []
for quad in poly_quads:
quad_h = (np.linalg.norm(quad[0] - quad[3]) + np.linalg.norm(quad[2] - quad[1])) / 2.0
quad_h = (np.linalg.norm(quad[0] - quad[3]) +
np.linalg.norm(quad[2] - quad[1])) / 2.0
height_list.append(quad_h)
average_height = max(sum(height_list) / len(height_list), 1.0)
return average_height
def generate_tcl_label(self, hw, polys, tags, ds_ratio,
tcl_ratio=0.3, shrink_ratio_of_width=0.15):
def generate_tcl_label(self,
hw,
polys,
tags,
ds_ratio,
tcl_ratio=0.3,
shrink_ratio_of_width=0.15):
"""
Generate polygon.
"""
......@@ -225,10 +251,17 @@ class SASTProcessTrain(object):
h, w = int(h * ds_ratio), int(w * ds_ratio)
polys = polys * ds_ratio
score_map = np.zeros((h, w,), dtype=np.float32)
score_map = np.zeros(
(
h,
w, ), dtype=np.float32)
tbo_map = np.zeros((h, w, 5), dtype=np.float32)
training_mask = np.ones((h, w,), dtype=np.float32)
direction_map = np.ones((h, w, 3)) * np.array([0, 0, 1]).reshape([1, 1, 3]).astype(np.float32)
training_mask = np.ones(
(
h,
w, ), dtype=np.float32)
direction_map = np.ones((h, w, 3)) * np.array([0, 0, 1]).reshape(
[1, 1, 3]).astype(np.float32)
for poly_idx, poly_tag in enumerate(zip(polys, tags)):
poly = poly_tag[0]
......@@ -236,9 +269,11 @@ class SASTProcessTrain(object):
# generate min_area_quad
min_area_quad, center_point = self.gen_min_area_quad_from_poly(poly)
min_area_quad_h = 0.5 * (np.linalg.norm(min_area_quad[0] - min_area_quad[3]) +
min_area_quad_h = 0.5 * (
np.linalg.norm(min_area_quad[0] - min_area_quad[3]) +
np.linalg.norm(min_area_quad[1] - min_area_quad[2]))
min_area_quad_w = 0.5 * (np.linalg.norm(min_area_quad[0] - min_area_quad[1]) +
min_area_quad_w = 0.5 * (
np.linalg.norm(min_area_quad[0] - min_area_quad[1]) +
np.linalg.norm(min_area_quad[2] - min_area_quad[3]))
if min(min_area_quad_h, min_area_quad_w) < self.min_text_size * ds_ratio \
......@@ -247,25 +282,37 @@ class SASTProcessTrain(object):
if tag:
# continue
cv2.fillPoly(training_mask, poly.astype(np.int32)[np.newaxis, :, :], 0.15)
cv2.fillPoly(training_mask,
poly.astype(np.int32)[np.newaxis, :, :], 0.15)
else:
tcl_poly = self.poly2tcl(poly, tcl_ratio)
tcl_quads = self.poly2quads(tcl_poly)
poly_quads = self.poly2quads(poly)
# stcl map
stcl_quads, quad_index = self.shrink_poly_along_width(tcl_quads, shrink_ratio_of_width=shrink_ratio_of_width,
stcl_quads, quad_index = self.shrink_poly_along_width(
tcl_quads,
shrink_ratio_of_width=shrink_ratio_of_width,
expand_height_ratio=1.0 / tcl_ratio)
# generate tcl map
cv2.fillPoly(score_map, np.round(stcl_quads).astype(np.int32), 1.0)
cv2.fillPoly(score_map,
np.round(stcl_quads).astype(np.int32), 1.0)
# generate tbo map
for idx, quad in enumerate(stcl_quads):
quad_mask = np.zeros((h, w), dtype=np.float32)
quad_mask = cv2.fillPoly(quad_mask, np.round(quad[np.newaxis, :, :]).astype(np.int32), 1.0)
tbo_map = self.gen_quad_tbo(poly_quads[quad_index[idx]], quad_mask, tbo_map)
quad_mask = cv2.fillPoly(
quad_mask,
np.round(quad[np.newaxis, :, :]).astype(np.int32), 1.0)
tbo_map = self.gen_quad_tbo(poly_quads[quad_index[idx]],
quad_mask, tbo_map)
return score_map, tbo_map, training_mask
def generate_tvo_and_tco(self, hw, polys, tags, tcl_ratio=0.3, ds_ratio=0.25):
def generate_tvo_and_tco(self,
hw,
polys,
tags,
tcl_ratio=0.3,
ds_ratio=0.25):
"""
Generate tcl map, tvo map and tbo map.
"""
......@@ -297,9 +344,11 @@ class SASTProcessTrain(object):
# generate min_area_quad
min_area_quad, center_point = self.gen_min_area_quad_from_poly(poly)
min_area_quad_h = 0.5 * (np.linalg.norm(min_area_quad[0] - min_area_quad[3]) +
min_area_quad_h = 0.5 * (
np.linalg.norm(min_area_quad[0] - min_area_quad[3]) +
np.linalg.norm(min_area_quad[1] - min_area_quad[2]))
min_area_quad_w = 0.5 * (np.linalg.norm(min_area_quad[0] - min_area_quad[1]) +
min_area_quad_w = 0.5 * (
np.linalg.norm(min_area_quad[0] - min_area_quad[1]) +
np.linalg.norm(min_area_quad[2] - min_area_quad[3]))
# generate tcl map and text, 128 * 128
......@@ -307,25 +356,32 @@ class SASTProcessTrain(object):
# generate poly_tv_xy_map
for idx in range(4):
cv2.fillPoly(poly_tv_xy_map[2 * idx],
cv2.fillPoly(
poly_tv_xy_map[2 * idx],
np.round(tcl_poly[np.newaxis, :, :]).astype(np.int32),
float(min(max(min_area_quad[idx, 0], 0), w)))
cv2.fillPoly(poly_tv_xy_map[2 * idx + 1],
cv2.fillPoly(
poly_tv_xy_map[2 * idx + 1],
np.round(tcl_poly[np.newaxis, :, :]).astype(np.int32),
float(min(max(min_area_quad[idx, 1], 0), h)))
# generate poly_tc_xy_map
for idx in range(2):
cv2.fillPoly(poly_tc_xy_map[idx],
np.round(tcl_poly[np.newaxis, :, :]).astype(np.int32), float(center_point[idx]))
cv2.fillPoly(
poly_tc_xy_map[idx],
np.round(tcl_poly[np.newaxis, :, :]).astype(np.int32),
float(center_point[idx]))
# generate poly_short_edge_map
cv2.fillPoly(poly_short_edge_map,
cv2.fillPoly(
poly_short_edge_map,
np.round(tcl_poly[np.newaxis, :, :]).astype(np.int32),
float(max(min(min_area_quad_h, min_area_quad_w), 1.0)))
# generate poly_mask and training_mask
cv2.fillPoly(poly_mask, np.round(tcl_poly[np.newaxis, :, :]).astype(np.int32), 1)
cv2.fillPoly(poly_mask,
np.round(tcl_poly[np.newaxis, :, :]).astype(np.int32),
1)
tvo_map *= poly_mask
tvo_map[:8] -= poly_tv_xy_map
......@@ -356,7 +412,8 @@ class SASTProcessTrain(object):
elif point_num > 4:
vector_1 = poly[0] - poly[1]
vector_2 = poly[1] - poly[2]
cos_theta = np.dot(vector_1, vector_2) / (np.linalg.norm(vector_1) * np.linalg.norm(vector_2) + 1e-6)
cos_theta = np.dot(vector_1, vector_2) / (
np.linalg.norm(vector_1) * np.linalg.norm(vector_2) + 1e-6)
theta = np.arccos(np.round(cos_theta, decimals=4))
if abs(theta) > (70 / 180 * math.pi):
......@@ -374,7 +431,8 @@ class SASTProcessTrain(object):
min_area_quad = poly
center_point = np.sum(poly, axis=0) / 4
else:
rect = cv2.minAreaRect(poly.astype(np.int32)) # (center (x,y), (width, height), angle of rotation)
rect = cv2.minAreaRect(poly.astype(
np.int32)) # (center (x,y), (width, height), angle of rotation)
center_point = rect[0]
box = np.array(cv2.boxPoints(rect))
......@@ -394,16 +452,23 @@ class SASTProcessTrain(object):
return min_area_quad, center_point
def shrink_quad_along_width(self, quad, begin_width_ratio=0., end_width_ratio=1.):
def shrink_quad_along_width(self,
quad,
begin_width_ratio=0.,
end_width_ratio=1.):
"""
Generate shrink_quad_along_width.
"""
ratio_pair = np.array([[begin_width_ratio], [end_width_ratio]], dtype=np.float32)
ratio_pair = np.array(
[[begin_width_ratio], [end_width_ratio]], dtype=np.float32)
p0_1 = quad[0] + (quad[1] - quad[0]) * ratio_pair
p3_2 = quad[3] + (quad[2] - quad[3]) * ratio_pair
return np.array([p0_1[0], p0_1[1], p3_2[1], p3_2[0]])
def shrink_poly_along_width(self, quads, shrink_ratio_of_width, expand_height_ratio=1.0):
def shrink_poly_along_width(self,
quads,
shrink_ratio_of_width,
expand_height_ratio=1.0):
"""
shrink poly with given length.
"""
......@@ -421,22 +486,28 @@ class SASTProcessTrain(object):
upper_edge_list.append(upper_edge_len)
# length of left edge and right edge.
left_length = np.linalg.norm(quads[0][0] - quads[0][3]) * expand_height_ratio
right_length = np.linalg.norm(quads[-1][1] - quads[-1][2]) * expand_height_ratio
left_length = np.linalg.norm(quads[0][0] - quads[0][
3]) * expand_height_ratio
right_length = np.linalg.norm(quads[-1][1] - quads[-1][
2]) * expand_height_ratio
shrink_length = min(left_length, right_length, sum(upper_edge_list)) * shrink_ratio_of_width
shrink_length = min(left_length, right_length,
sum(upper_edge_list)) * shrink_ratio_of_width
# shrinking length
upper_len_left = shrink_length
upper_len_right = sum(upper_edge_list) - shrink_length
left_idx, left_ratio = get_cut_info(upper_edge_list, upper_len_left)
left_quad = self.shrink_quad_along_width(quads[left_idx], begin_width_ratio=left_ratio, end_width_ratio=1)
left_quad = self.shrink_quad_along_width(
quads[left_idx], begin_width_ratio=left_ratio, end_width_ratio=1)
right_idx, right_ratio = get_cut_info(upper_edge_list, upper_len_right)
right_quad = self.shrink_quad_along_width(quads[right_idx], begin_width_ratio=0, end_width_ratio=right_ratio)
right_quad = self.shrink_quad_along_width(
quads[right_idx], begin_width_ratio=0, end_width_ratio=right_ratio)
out_quad_list = []
if left_idx == right_idx:
out_quad_list.append([left_quad[0], right_quad[1], right_quad[2], left_quad[3]])
out_quad_list.append(
[left_quad[0], right_quad[1], right_quad[2], left_quad[3]])
else:
out_quad_list.append(left_quad)
for idx in range(left_idx + 1, right_idx):
......@@ -500,7 +571,8 @@ class SASTProcessTrain(object):
"""
Generate center line by poly clock-wise point. (4, 2)
"""
ratio_pair = np.array([[0.5 - ratio / 2], [0.5 + ratio / 2]], dtype=np.float32)
ratio_pair = np.array(
[[0.5 - ratio / 2], [0.5 + ratio / 2]], dtype=np.float32)
p0_3 = poly[0] + (poly[3] - poly[0]) * ratio_pair
p1_2 = poly[1] + (poly[2] - poly[1]) * ratio_pair
return np.array([p0_3[0], p1_2[0], p1_2[1], p0_3[1]])
......@@ -509,12 +581,14 @@ class SASTProcessTrain(object):
"""
Generate center line by poly clock-wise point.
"""
ratio_pair = np.array([[0.5 - ratio / 2], [0.5 + ratio / 2]], dtype=np.float32)
ratio_pair = np.array(
[[0.5 - ratio / 2], [0.5 + ratio / 2]], dtype=np.float32)
tcl_poly = np.zeros_like(poly)
point_num = poly.shape[0]
for idx in range(point_num // 2):
point_pair = poly[idx] + (poly[point_num - 1 - idx] - poly[idx]) * ratio_pair
point_pair = poly[idx] + (poly[point_num - 1 - idx] - poly[idx]
) * ratio_pair
tcl_poly[idx] = point_pair[0]
tcl_poly[point_num - 1 - idx] = point_pair[1]
return tcl_poly
......@@ -527,8 +601,10 @@ class SASTProcessTrain(object):
up_line = self.line_cross_two_point(quad[0], quad[1])
lower_line = self.line_cross_two_point(quad[3], quad[2])
quad_h = 0.5 * (np.linalg.norm(quad[0] - quad[3]) + np.linalg.norm(quad[1] - quad[2]))
quad_w = 0.5 * (np.linalg.norm(quad[0] - quad[1]) + np.linalg.norm(quad[2] - quad[3]))
quad_h = 0.5 * (np.linalg.norm(quad[0] - quad[3]) +
np.linalg.norm(quad[1] - quad[2]))
quad_w = 0.5 * (np.linalg.norm(quad[0] - quad[1]) +
np.linalg.norm(quad[2] - quad[3]))
# average angle of left and right line.
angle = self.average_angle(quad)
......@@ -565,7 +641,8 @@ class SASTProcessTrain(object):
quad_num = point_num // 2 - 1
for idx in range(quad_num):
# reshape and adjust to clock-wise
quad_list.append((np.array(point_pair_list)[[idx, idx + 1]]).reshape(4, 2)[[0, 2, 3, 1]])
quad_list.append((np.array(point_pair_list)[[idx, idx + 1]]
).reshape(4, 2)[[0, 2, 3, 1]])
return np.array(quad_list)
......@@ -579,7 +656,8 @@ class SASTProcessTrain(object):
return None
h, w, _ = im.shape
text_polys, text_tags, hv_tags = self.check_and_validate_polys(text_polys, text_tags, (h, w))
text_polys, text_tags, hv_tags = self.check_and_validate_polys(
text_polys, text_tags, (h, w))
if text_polys.shape[0] == 0:
return None
......@@ -621,7 +699,8 @@ class SASTProcessTrain(object):
return None
#resize image
std_ratio = float(self.input_size) / max(new_w, new_h)
rand_scales = np.array([0.25, 0.375, 0.5, 0.625, 0.75, 0.875, 1.0, 1.0, 1.0, 1.0, 1.0])
rand_scales = np.array(
[0.25, 0.375, 0.5, 0.625, 0.75, 0.875, 1.0, 1.0, 1.0, 1.0, 1.0])
rz_scale = std_ratio * np.random.choice(rand_scales)
im = cv2.resize(im, dsize=None, fx=rz_scale, fy=rz_scale)
text_polys[:, :, 0] *= rz_scale
......@@ -630,7 +709,7 @@ class SASTProcessTrain(object):
#add gaussian blur
if np.random.rand() < 0.1 * 0.5:
ks = np.random.permutation(5)[0] + 1
ks = int(ks/2)*2 + 1
ks = int(ks / 2) * 2 + 1
im = cv2.GaussianBlur(im, ksize=(ks, ks), sigmaX=0, sigmaY=0)
#add brighter
if np.random.rand() < 0.1 * 0.5:
......@@ -646,7 +725,8 @@ class SASTProcessTrain(object):
if min(new_w, new_h) < self.input_size * 0.5:
return None
im_padded = np.ones((self.input_size, self.input_size, 3), dtype=np.float32)
im_padded = np.ones(
(self.input_size, self.input_size, 3), dtype=np.float32)
im_padded[:, :, 2] = 0.485 * 255
im_padded[:, :, 1] = 0.456 * 255
im_padded[:, :, 0] = 0.406 * 255
......@@ -661,15 +741,20 @@ class SASTProcessTrain(object):
sw = int(np.random.rand() * del_w)
# Padding
im_padded[sh: sh + new_h, sw: sw + new_w, :] = im.copy()
im_padded[sh:sh + new_h, sw:sw + new_w, :] = im.copy()
text_polys[:, :, 0] += sw
text_polys[:, :, 1] += sh
score_map, border_map, training_mask = self.generate_tcl_label((self.input_size, self.input_size),
text_polys, text_tags, 0.25)
score_map, border_map, training_mask = self.generate_tcl_label(
(self.input_size, self.input_size), text_polys, text_tags, 0.25)
# SAST head
tvo_map, tco_map = self.generate_tvo_and_tco((self.input_size, self.input_size), text_polys, text_tags, tcl_ratio=0.3, ds_ratio=0.25)
tvo_map, tco_map = self.generate_tvo_and_tco(
(self.input_size, self.input_size),
text_polys,
text_tags,
tcl_ratio=0.3,
ds_ratio=0.25)
# print("test--------tvo_map shape:", tvo_map.shape)
im_padded[:, :, 2] -= 0.485 * 255
......
ppocr/data/lmdb_dataset.py
浏览文件 @ 5e9fb50d
......@@ -21,7 +21,7 @@ from .imaug import transform, create_operators
class LMDBDateSet(Dataset):
def __init__(self, config, mode, logger):
def __init__(self, config, mode, logger, seed=None):
super(LMDBDateSet, self).__init__()
global_config = config['Global']
......
ppocr/data/simple_dataset.py
浏览文件 @ 5e9fb50d
......@@ -20,7 +20,7 @@ from .imaug import transform, create_operators
class SimpleDataSet(Dataset):
def __init__(self, config, mode, logger):
def __init__(self, config, mode, logger, seed=None):
super(SimpleDataSet, self).__init__()
self.logger = logger
......@@ -41,6 +41,7 @@ class SimpleDataSet(Dataset):
self.data_dir = dataset_config['data_dir']
self.do_shuffle = loader_config['shuffle']
self.seed = seed
logger.info("Initialize indexs of datasets:%s" % label_file_list)
self.data_lines = self.get_image_info_list(label_file_list, ratio_list)
self.data_idx_order_list = list(range(len(self.data_lines)))
......@@ -55,6 +56,7 @@ class SimpleDataSet(Dataset):
for idx, file in enumerate(file_list):
with open(file, "rb") as f:
lines = f.readlines()
random.seed(self.seed)
lines = random.sample(lines,
round(len(lines) * ratio_list[idx]))
data_lines.extend(lines)
......@@ -62,6 +64,7 @@ class SimpleDataSet(Dataset):
def shuffle_data_random(self):
if self.do_shuffle:
random.seed(self.seed)
random.shuffle(self.data_lines)
return
......
ppocr/modeling/transforms/tps.py
浏览文件 @ 5e9fb50d
......@@ -213,16 +213,14 @@ class GridGenerator(nn.Layer):
def build_P_paddle(self, I_r_size):
I_r_height, I_r_width = I_r_size
I_r_grid_x = paddle.divide(
paddle.arange(
-I_r_width, I_r_width, 2, dtype='float64') + 1.0,
paddle.to_tensor(
I_r_width, dtype='float64'))
I_r_grid_y = paddle.divide(
paddle.arange(
-I_r_height, I_r_height, 2, dtype='float64') + 1.0,
paddle.to_tensor(
I_r_height, dtype='float64')) # self.I_r_height
I_r_grid_x = (paddle.arange(
-I_r_width, I_r_width, 2, dtype='float64') + 1.0
) / paddle.to_tensor(np.array([I_r_width]))
I_r_grid_y = (paddle.arange(
-I_r_height, I_r_height, 2, dtype='float64') + 1.0
) / paddle.to_tensor(np.array([I_r_height]))
# P: self.I_r_width x self.I_r_height x 2
P = paddle.stack(paddle.meshgrid(I_r_grid_x, I_r_grid_y), axis=2)
P = paddle.transpose(P, perm=[1, 0, 2])
......
ppocr/postprocess/rec_postprocess.py
浏览文件 @ 5e9fb50d
......@@ -109,7 +109,7 @@ class CTCLabelDecode(BaseRecLabelDecode):
preds_idx = preds.argmax(axis=2)
preds_prob = preds.max(axis=2)
text = self.decode(preds_idx, preds_prob)
text = self.decode(preds_idx, preds_prob, is_remove_duplicate=True)
if label is None:
return text
label = self.decode(label)
......
tools/program.py
浏览文件 @ 5e9fb50d
......@@ -182,8 +182,8 @@ def train(config,
start_epoch = 1
for epoch in range(start_epoch, epoch_num + 1):
if epoch > 0:
train_dataloader = build_dataloader(config, 'Train', device, logger)
train_dataloader = build_dataloader(
config, 'Train', device, logger, seed=epoch)
train_batch_cost = 0.0
train_reader_cost = 0.0
batch_sum = 0
......
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