提交 721c76b4 编写于 作者: L LDOUBLEV

fix conflict

---
name: Issue template
about: Issue template for code error.
title: ''
labels: ''
assignees: ''
---
请提供下述完整信息以便快速定位问题/Please provide the following information to quickly locate the problem
- 系统环境/System Environment:
- 版本号/Version:Paddle: PaddleOCR: 问题相关组件/Related components:
- 运行指令/Command Code:
- 完整报错/Complete Error Message:
......@@ -5,5 +5,6 @@ recursive-include ppocr/utils *.txt utility.py logging.py network.py
recursive-include ppocr/data *.py
recursive-include ppocr/postprocess *.py
recursive-include tools/infer *.py
recursive-include tools __init__.py
recursive-include ppocr/utils/e2e_utils *.py
recursive-include ppstructure *.py
\ No newline at end of file
......@@ -35,6 +35,7 @@ import numpy as np
sys.path.append(__dir__)
sys.path.append(os.path.abspath(os.path.join(__dir__, '../..')))
sys.path.append(os.path.abspath(os.path.join(__dir__, '../PaddleOCR')))
sys.path.append("..")
from paddleocr import PaddleOCR
......@@ -113,7 +114,7 @@ class MainWindow(QMainWindow, WindowMixin):
getStr = lambda strId: self.stringBundle.getString(strId)
self.defaultSaveDir = defaultSaveDir
self.ocr = PaddleOCR(use_pdserving=False, use_angle_cls=True, det=True, cls=True, use_gpu=gpu, lang=lang)
self.ocr = PaddleOCR(use_pdserving=False, use_angle_cls=True, det=True, cls=True, use_gpu=gpu, lang=lang, show_log=False)
if os.path.exists('./data/paddle.png'):
result = self.ocr.ocr('./data/paddle.png', cls=True, det=True)
......@@ -390,7 +391,7 @@ class MainWindow(QMainWindow, WindowMixin):
'Ctrl+J', 'edit', u'Move and edit Boxs', enabled=False)
create = action(getStr('crtBox'), self.createShape,
'w', 'new', getStr('crtBoxDetail'), enabled=False)
'w', 'objects', getStr('crtBoxDetail'), enabled=False)
delete = action(getStr('delBox'), self.deleteSelectedShape,
'backspace', 'delete', getStr('delBoxDetail'), enabled=False)
......@@ -1388,7 +1389,6 @@ class MainWindow(QMainWindow, WindowMixin):
for box in self.PPlabel[imgidx]:
shapes.append((box['transcription'], box['points'], None, None, box['difficult']))
print(shapes)
self.loadLabels(shapes)
self.canvas.verified = False
......
......@@ -8,7 +8,10 @@ PPOCRLabel is a semi-automatic graphic annotation tool suitable for OCR field, w
### Recent Update
- 2021.8.11:
- 2021.11.17:
- Support install and start PPOCRLabel through the whl package (by [d2623587501](https://github.com/d2623587501))
- Dataset segmentation: Divide the annotation file into training, verification and testing parts (refer to section 3.5 below, by [MrCuiHao](https://github.com/MrCuiHao))
- 2021.8.11:
- New functions: Open the dataset folder, image rotation (Note: Please delete the label box before rotating the image) (by [Wei-JL](https://github.com/Wei-JL))
- Added shortcut key description (Help-Shortcut Key), repaired the direction shortcut key movement function under batch processing (by [d2623587501](https://github.com/d2623587501))
- 2021.2.5: New batch processing and undo functions (by [Evezerest](https://github.com/Evezerest)):
......@@ -21,14 +24,11 @@ PPOCRLabel is a semi-automatic graphic annotation tool suitable for OCR field, w
- Click to modify the recognition result.(If you can't change the result, please switch to the system default input method, or switch back to the original input method again)
- 2020.12.18: Support re-recognition of a single label box (by [ninetailskim](https://github.com/ninetailskim) ), perfect shortcut keys.
### TODO:
- Lock box mode: For the same scene data, the size and position of the locked detection box can be transferred between different pictures.
## Installation
### 1. Environment Preparation
## 1. Installation and Run
#### **Install PaddlePaddle 2.0**
### 1.1 Install PaddlePaddle
```bash
pip3 install --upgrade pip
......@@ -42,61 +42,57 @@ 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.
#### **Install PaddleOCR**
### 1.2 Install and Run PPOCRLabel
```bash
# Recommend
git clone https://github.com/PaddlePaddle/PaddleOCR
# If you cannot pull successfully due to network problems, you can also choose to use the code hosting on the cloud:
git clone https://gitee.com/paddlepaddle/PaddleOCR
# Note: The cloud-hosting code may not be able to synchronize the update with this GitHub project in real time. There might be a delay of 3-5 days. Please give priority to the recommended method.
```
PPOCRLabel can be started in two ways: whl package and Python script. The whl package form is more convenient to start, and the python script to start is convenient for secondary development.
#### **Install Third-party Libraries**
#### Windows
```bash
cd PaddleOCR
pip3 install -r requirements.txt
pip install PPOCRLabel # install
PPOCRLabel # run
```
If you getting this error `OSError: [WinError 126] The specified module could not be found` when you install shapely on windows. Please try to download Shapely whl file using http://www.lfd.uci.edu/~gohlke/pythonlibs/#shapely.
Reference: [Solve shapely installation on windows](https://stackoverflow.com/questions/44398265/install-shapely-oserror-winerror-126-the-specified-module-could-not-be-found)
> If you getting this error `OSError: [WinError 126] The specified module could not be found` when you install shapely on windows. Please try to download Shapely whl file using http://www.lfd.uci.edu/~gohlke/pythonlibs/#shapely.
>
> Reference: [Solve shapely installation on windows](https://stackoverflow.com/questions/44398265/install-shapely-oserror-winerror-126-the-specified-module-could-not-be-found)
>
### 2. Install PPOCRLabel
#### Ubuntu Linux
#### Windows
```bash
pip3 install PPOCRLabel
pip3 install trash-cli
PPOCRLabel
```
#### MacOS
```bash
pip install pyqt5
cd ./PPOCRLabel # Change the directory to the PPOCRLabel folder
python PPOCRLabel.py
pip3 install PPOCRLabel
pip3 install opencv-contrib-python-headless==4.2.0.32
PPOCRLabel # run
```
#### Ubuntu Linux
#### 1.2.2 Build and Install the Whl Package Locally
```bash
pip3 install pyqt5
pip3 install trash-cli
cd ./PPOCRLabel # Change the directory to the PPOCRLabel folder
python3 PPOCRLabel.py
cd PaddleOCR/PPOCRLabel
python3 setup.py bdist_wheel
pip3 install dist/PPOCRLabel-1.0.0-py2.py3-none-any.whl
```
#### MacOS
#### 1.2.3 Run PPOCRLabel by Python Script
```bash
pip3 install pyqt5
pip3 uninstall opencv-python # Uninstall opencv manually as it conflicts with pyqt
pip3 install opencv-contrib-python-headless==4.2.0.32 # Install the headless version of opencv
cd ./PPOCRLabel # Change the directory to the PPOCRLabel folder
python3 PPOCRLabel.py
cd ./PPOCRLabel # Switch to the PPOCRLabel directory
python PPOCRLabel.py --lang ch
```
## Usage
### Steps
## 2. Usage
### 2.1 Steps
1. Build and launch using the instructions above.
......@@ -122,7 +118,7 @@ python3 PPOCRLabel.py
10. Labeling result: the user can export the label result manually through the menu "File - Export Label", while the program will also export automatically if "File - Auto export Label Mode" is selected. The manually checked label will be stored in *Label.txt* under the opened picture folder. Click "File"-"Export Recognition Results" in the menu bar, the recognition training data of such pictures will be saved in the *crop_img* folder, and the recognition label will be saved in *rec_gt.txt*<sup>[4]</sup>.
### Note
### 2.2 Note
[1] PPOCRLabel uses the opened folder as the project. After opening the image folder, the picture will not be displayed in the dialog. Instead, the pictures under the folder will be directly imported into the program after clicking "Open Dir".
......@@ -140,9 +136,11 @@ python3 PPOCRLabel.py
| rec_gt.txt | The recognition label file, which can be directly used for PPOCR identification model training, is generated after the user clicks on the menu bar "File"-"Export recognition result". |
| crop_img | The recognition data, generated at the same time with *rec_gt.txt* |
## Explanation
### Shortcut keys
## 3. Explanation
### 3.1 Shortcut keys
| Shortcut keys | Description |
| ------------------------ | ------------------------------------------------ |
......@@ -162,31 +160,56 @@ python3 PPOCRLabel.py
| Ctrl-- | Zoom out |
| ↑→↓← | Move selected box |
### Built-in Model
### 3.2 Built-in Model
- Default model: PPOCRLabel uses the Chinese and English ultra-lightweight OCR model in PaddleOCR by default, supports Chinese, English and number recognition, and multiple language detection.
- Model language switching: Changing the built-in model language is supportable by clicking "PaddleOCR"-"Choose OCR Model" in the menu bar. Currently supported languages​include French, German, Korean, and Japanese.
For specific model download links, please refer to [PaddleOCR Model List](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/doc/doc_en/models_list_en.md#multilingual-recognition-modelupdating)
- Custom model: The model trained by users can be replaced by modifying PPOCRLabel.py in [PaddleOCR class instantiation](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/PPOCRLabel/PPOCRLabel.py#L110) referring [Custom Model Code](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/doc/doc_en/whl_en.md#use-custom-model)
- **Custom Model**: If users want to replace the built-in model with their own inference model, they can follow the [Custom Model Code Usage](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.3/doc/doc_en/whl_en.md#31-use-by-code) by modifying PPOCRLabel.py for [Instantiation of PaddleOCR class](https://github.com/PaddlePaddle/PaddleOCR/blob/release/ 2.3/PPOCRLabel/PPOCRLabel.py#L116) :
### Export Label Result
add parameter `det_model_dir` in `self.ocr = PaddleOCR(use_pdserving=False, use_angle_cls=True, det=True, cls=True, use_gpu=gpu, lang=lang) `
### 3.3 Export Label Result
PPOCRLabel supports three ways to export Label.txt
- Automatically export: After selecting "File - Auto Export Label Mode", the program will automatically write the annotations into Label.txt every time the user confirms an image. If this option is not turned on, it will be automatically exported after detecting that the user has manually checked 5 images.
> The automatically export mode is turned off by default
- Manual export: Click "File-Export Marking Results" to manually export the label.
- Close application export
### Export Partial Recognition Results
### 3.4 Export Partial Recognition Results
For some data that are difficult to recognize, the recognition results will not be exported by **unchecking** the corresponding tags in the recognition results checkbox. The unchecked recognition result is saved as `True` in the `difficult` variable in the label file `label.txt`.
> *Note: The status of the checkboxes in the recognition results still needs to be saved manually by clicking Save Button.*
For some data that are difficult to recognize, the recognition results will not be exported by **unchecking** the corresponding tags in the recognition results checkbox.
### 3.5 Dataset division
*Note: The status of the checkboxes in the recognition results still needs to be saved manually by clicking Save Button.*
- Enter the following command in the terminal to execute the dataset division script:
### Error message
```
cd ./PPOCRLabel # Change the directory to the PPOCRLabel folder
python gen_ocr_train_val_test.py --trainValTestRatio 6:2:2 --labelRootPath ../train_data/label --detRootPath ../train_data/det --recRootPath ../train_data/rec
```
Parameter Description:
- `trainValTestRatio` is the division ratio of the number of images in the training set, validation set, and test set, set according to your actual situation, the default is `6:2:2`
- `labelRootPath` is the storage path of the dataset labeled by PPOCRLabel, the default is `../train_data/label`
- `detRootPath` is the path where the text detection dataset is divided according to the dataset marked by PPOCRLabel. The default is `../train_data/det`
- `recRootPath` is the path where the character recognition dataset is divided according to the dataset marked by PPOCRLabel. The default is `../train_data/rec`
### 3.6 Error message
- If paddleocr is installed with whl, it has a higher priority than calling PaddleOCR class with paddleocr.py, which may cause an exception if whl package is not updated.
......@@ -204,6 +227,8 @@ For some data that are difficult to recognize, the recognition results will not
pip install opencv-contrib-python-headless==4.2.0.32
```
### Related
### 4. Related
1.[Tzutalin. LabelImg. Git code (2015)](https://github.com/tzutalin/labelImg)
\ No newline at end of file
......@@ -2,14 +2,17 @@
# PPOCRLabel
PPOCRLabel是一款适用于OCR领域的半自动化图形标注工具,内置PPOCR模型对数据自动标注和重新识别。使用python3和pyqt5编写,支持矩形框标注和四点标注模式,导出格式可直接用于PPOCR检测和识别模型的训练。
PPOCRLabel是一款适用于OCR领域的半自动化图形标注工具,内置PP-OCR模型对数据自动标注和重新识别。使用Python3和PyQT5编写,支持矩形框标注和四点标注模式,导出格式可直接用于PaddleOCR检测和识别模型的训练。
<img src="./data/gif/steps.gif" width="100%"/>
#### 近期更新
- 2021.11.17:
- 新增支持通过whl包安装和启动PPOCRLabel(by [d2623587501](https://github.com/d2623587501)
- 标注数据集切分:对标注数据进行训练、验证与测试集划分(参考下方3.5节,by [MrCuiHao](https://github.com/MrCuiHao)
- 2021.8.11:
- 新增功能:打开数据所在文件夹、图像旋转(注意:旋转前的图片上不能存在标记框)(by [Wei-JL](https://github.com/Wei-JL)
- 新增功能:打开数据所在文件夹、右键图像旋转90度(注意:旋转前的图片上不能存在标记框,by [Wei-JL](https://github.com/Wei-JL)
- 新增快捷键说明(帮助-快捷键)、修复批处理下的方向快捷键移动功能(by [d2623587501](https://github.com/d2623587501)
- 2021.2.5:新增批处理与撤销功能(by [Evezerest](https://github.com/Evezerest))
- **批处理功能**:按住Ctrl键选择标记框后可批量移动、复制、删除、重新识别。
......@@ -21,85 +24,75 @@ PPOCRLabel是一款适用于OCR领域的半自动化图形标注工具,内置P
- 识别结果更改为单击修改。(如果无法修改,请切换为系统自带输入法,或再次切回原输入法)
- 2020.12.18: 支持对单个标记框进行重新识别(by [ninetailskim](https://github.com/ninetailskim)),完善快捷键。
#### 尽请期待
- 锁定框模式:针对同一场景数据,被锁定的检测框的大小与位置能在不同图片之间传递。
如果您对以上内容感兴趣或对完善工具有不一样的想法,欢迎加入我们的SIG队伍与我们共同开发。可以在[此处](https://github.com/PaddlePaddle/PaddleOCR/issues/1728)完成问卷和前置任务,经过我们确认相关内容后即可正式加入,享受SIG福利,共同为OCR开源事业贡献(特别说明:针对PPOCRLabel的改进也属于PaddleOCR前置任务)
## 安装
### 1. 环境搭建
#### 安装PaddlePaddle
## 1. 安装与运行
### 1.1 安装PaddlePaddle
```bash
pip3 install --upgrade pip
如果您的机器安装的是CUDA9或CUDA10,请运行以下命令安装
# 如果您的机器安装的是CUDA9或CUDA10,请运行以下命令安装
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
# 如果您的机器是CPU,请运行以下命令安装
python3 -m pip install paddlepaddle -i https://mirror.baidu.com/pypi/simple
```
更多的版本需求,请参照[安装文档](https://www.paddlepaddle.org.cn/install/quick)中的说明进行操作。
#### **安装PaddleOCR**
### 1.2 安装与运行PPOCRLabel
```bash
【推荐】git clone https://github.com/PaddlePaddle/PaddleOCR
PPOCRLabel可通过whl包与Python脚本两种方式启动,whl包形式启动更加方便,python脚本启动便于二次开发
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:
#### 1.2.1 通过whl包安装与运行
git clone https://gitee.com/paddlepaddle/PaddleOCR
##### Windows
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。
```bash
pip install PPOCRLabel # 安装
PPOCRLabel --lang ch # 运行
```
#### 安装第三方库
> 注意:通过whl包安装PPOCRLabel会自动下载 `paddleocr` whl包,其中shapely依赖可能会出现 `[winRrror 126] 找不到指定模块的问题。` 的错误,建议从[这里](https://www.lfd.uci.edu/~gohlke/pythonlibs/#shapely)下载并安装
##### Ubuntu Linux
```bash
cd PaddleOCR
pip3 install -r requirements.txt
pip3 install PPOCRLabel
pip3 install trash-cli
PPOCRLabel --lang ch
```
注意,windows环境下,建议从[这里](https://www.lfd.uci.edu/~gohlke/pythonlibs/#shapely)下载shapely安装包完成安装, 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`
### 2. 安装PPOCRLabel
#### Windows
##### MacOS
```bash
pip install pyqt5
cd ./PPOCRLabel # 将目录切换到PPOCRLabel文件夹下
python PPOCRLabel.py --lang ch
pip3 install PPOCRLabel
pip3 install opencv-contrib-python-headless==4.2.0.32 # 如果下载过慢请添加"-i https://mirror.baidu.com/pypi/simple"
PPOCRLabel --lang ch # 启动
```
#### Ubuntu Linux
#### 1.2.2 本地构建whl包并安装
```bash
pip3 install pyqt5
pip3 install trash-cli
cd ./PPOCRLabel # 将目录切换到PPOCRLabel文件夹下
python3 PPOCRLabel.py --lang ch
cd PaddleOCR/PPOCRLabel
python3 setup.py bdist_wheel
pip3 install dist/PPOCRLabel-1.0.0-py2.py3-none-any.whl -i https://mirror.baidu.com/pypi/simple
```
#### MacOS
#### 1.2.3 通过Python脚本运行PPOCRLabel
如果您对PPOCRLabel文件有所更改,通过Python脚本运行会更加方面的看到更改的结果
```bash
pip3 install pyqt5
pip3 uninstall opencv-python # 由于mac版本的opencv与pyqt有冲突,需先手动卸载opencv
pip3 install opencv-contrib-python-headless==4.2.0.32 # 安装headless版本的open-cv
cd ./PPOCRLabel # 将目录切换到PPOCRLabel文件夹下
python3 PPOCRLabel.py --lang ch
cd ./PPOCRLabel # 切换到PPOCRLabel目录
python PPOCRLabel.py --lang ch
```
## 使用
## 2. 使用
### 操作步骤
### 2.1 操作步骤
1. 安装与运行:使用上述命令安装与运行程序。
2. 打开文件夹:在菜单栏点击 “文件” - "打开目录" 选择待标记图片的文件夹<sup>[1]</sup>.
......@@ -112,7 +105,7 @@ python3 PPOCRLabel.py --lang ch
9. 删除:点击 “删除图像”,图片将会被删除至回收站。
10. 导出结果:用户可以通过菜单中“文件-导出标记结果”手动导出,同时也可以点击“文件 - 自动导出标记结果”开启自动导出。手动确认过的标记将会被存放在所打开图片文件夹下的*Label.txt*中。在菜单栏点击 “文件” - "导出识别结果"后,会将此类图片的识别训练数据保存在*crop_img*文件夹下,识别标签保存在*rec_gt.txt*<sup>[4]</sup>
### 注意
### 2.2 注意
[1] PPOCRLabel以文件夹为基本标记单位,打开待标记的图片文件夹后,不会在窗口栏中显示图片,而是在点击 "选择文件夹" 之后直接将文件夹下的图片导入到程序中。
......@@ -130,9 +123,11 @@ python3 PPOCRLabel.py --lang ch
| rec_gt.txt | 识别标签。可直接用于PPOCR识别模型训练。需用户手动点击菜单栏“文件” - "导出识别结果"后产生。 |
| crop_img | 识别数据。按照检测框切割后的图片。与rec_gt.txt同时产生。 |
## 说明
### 快捷键
## 3. 说明
### 3.1 快捷键
| 快捷键 | 说明 |
| ---------------- | ---------------------------- |
......@@ -152,29 +147,53 @@ python3 PPOCRLabel.py --lang ch
| Ctrl-- | 放大 |
| ↑→↓← | 移动标记框 |
### 内置模型
### 3.2 内置模型
- 默认模型:PPOCRLabel默认使用PaddleOCR中的中英文超轻量OCR模型,支持中英文与数字识别,多种语言检测。
- 模型语言切换:用户可通过菜单栏中 "PaddleOCR" - "选择模型" 切换内置模型语言,目前支持的语言包括法文、德文、韩文、日文。具体模型下载链接可参考[PaddleOCR模型列表](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/doc/doc_ch/models_list.md).
- 自定义模型:用户可根据[自定义模型代码使用](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/doc/doc_ch/whl.md#%E8%87%AA%E5%AE%9A%E4%B9%89%E6%A8%A1%E5%9E%8B),通过修改PPOCRLabel.py中针对[PaddleOCR类的实例化](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/PPOCRLabel/PPOCRLabel.py#L110)替换成自己训练的模型。
- **自定义模型**:如果用户想将内置模型更换为自己的推理模型,可根据[自定义模型代码使用](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/doc/doc_ch/whl.md#%E8%87%AA%E5%AE%9A%E4%B9%89%E6%A8%A1%E5%9E%8B),通过修改PPOCRLabel.py中针对[PaddleOCR类的实例化](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.3/PPOCRLabel/PPOCRLabel.py#L116) 实现,例如指定检测模型:`self.ocr = PaddleOCR(det=True, cls=True, use_gpu=gpu, lang=lang) `,在 `det_model_dir` 中传入 自己的模型即可。
### 导出标记结果
### 3.3 导出标记结果
PPOCRLabel支持三种导出方式:
- 自动导出:点击“文件 - 自动导出标记结果”后,用户每确认过一张图片,程序自动将标记结果写入Label.txt中。若未开启此选项,则检测到用户手动确认过5张图片后进行自动导出。
> 默认情况下自动导出功能为关闭状态
- 手动导出:点击“文件 - 导出标记结果”手动导出标记。
- 关闭应用程序导出
### 导出部分识别结果
### 3.4 导出部分识别结果
针对部分难以识别的数据,通过在识别结果的复选框中**取消勾选**相应的标记,其识别结果不会被导出。被取消勾选的识别结果在标记文件 `label.txt` 中的 `difficult` 变量保存为 `True`
> *注意:识别结果中的复选框状态仍需用户手动点击确认后才能保留*
### 3.5 数据集划分
在终端中输入以下命令执行数据集划分脚本:
```
cd ./PPOCRLabel # 将目录切换到PPOCRLabel文件夹下
python gen_ocr_train_val_test.py --trainValTestRatio 6:2:2 --labelRootPath ../train_data/label --detRootPath ../train_data/det --recRootPath ../train_data/rec
```
参数说明:
- `trainValTestRatio` 是训练集、验证集、测试集的图像数量划分比例,根据实际情况设定,默认是`6:2:2`
针对部分难以识别的数据,通过在识别结果的复选框中**取消勾选**相应的标记,其识别结果不会被导出。
- `labelRootPath` 是PPOCRLabel标注的数据集存放路径,默认是`../train_data/label`
*注意:识别结果中的复选框状态仍需用户手动点击确认后才能保留*
- `detRootPath` 是根据PPOCRLabel标注的数据集划分后的文本检测数据集存放的路径,默认是`../train_data/det `
- `recRootPath` 是根据PPOCRLabel标注的数据集划分后的字符识别数据集存放的路径,默认是`../train_data/rec`
### 3.6 错误提示
### 错误提示
- 如果同时使用whl包安装了paddleocr,其优先级大于通过paddleocr.py调用PaddleOCR类,whl包未更新时会导致程序异常。
- PPOCRLabel**不支持对中文文件名**的图片进行自动标注。
......@@ -194,6 +213,8 @@ PPOCRLabel支持三种导出方式:
pip install opencv-contrib-python-headless==4.2.0.32
```
### 参考资料
### 4. 参考资料
1.[Tzutalin. LabelImg. Git code (2015)](https://github.com/tzutalin/labelImg)
# coding:utf8
import os
import shutil
import random
import argparse
# 删除划分的训练集、验证集、测试集文件夹,重新创建一个空的文件夹
def isCreateOrDeleteFolder(path, flag):
flagPath = os.path.join(path, flag)
if os.path.exists(flagPath):
shutil.rmtree(flagPath)
os.makedirs(flagPath)
flagAbsPath = os.path.abspath(flagPath)
return flagAbsPath
def splitTrainVal(root, dir, absTrainRootPath, absValRootPath, absTestRootPath, trainTxt, valTxt, testTxt, flag):
# 按照指定的比例划分训练集、验证集、测试集
labelPath = os.path.join(root, dir)
labelAbsPath = os.path.abspath(labelPath)
if flag == "det":
labelFilePath = os.path.join(labelAbsPath, args.detLabelFileName)
elif flag == "rec":
labelFilePath = os.path.join(labelAbsPath, args.recLabelFileName)
labelFileRead = open(labelFilePath, "r", encoding="UTF-8")
labelFileContent = labelFileRead.readlines()
random.shuffle(labelFileContent)
labelRecordLen = len(labelFileContent)
for index, labelRecordInfo in enumerate(labelFileContent):
imageRelativePath = labelRecordInfo.split('\t')[0]
imageLabel = labelRecordInfo.split('\t')[1]
imageName = os.path.basename(imageRelativePath)
if flag == "det":
imagePath = os.path.join(labelAbsPath, imageName)
elif flag == "rec":
imagePath = os.path.join(labelAbsPath, "{}\\{}".format(args.recImageDirName, imageName))
# 按预设的比例划分训练集、验证集、测试集
trainValTestRatio = args.trainValTestRatio.split(":")
trainRatio = eval(trainValTestRatio[0]) / 10
valRatio = trainRatio + eval(trainValTestRatio[1]) / 10
curRatio = index / labelRecordLen
if curRatio < trainRatio:
imageCopyPath = os.path.join(absTrainRootPath, imageName)
shutil.copy(imagePath, imageCopyPath)
trainTxt.write("{}\t{}".format(imageCopyPath, imageLabel))
elif curRatio >= trainRatio and curRatio < valRatio:
imageCopyPath = os.path.join(absValRootPath, imageName)
shutil.copy(imagePath, imageCopyPath)
valTxt.write("{}\t{}".format(imageCopyPath, imageLabel))
else:
imageCopyPath = os.path.join(absTestRootPath, imageName)
shutil.copy(imagePath, imageCopyPath)
testTxt.write("{}\t{}".format(imageCopyPath, imageLabel))
# 删掉存在的文件
def removeFile(path):
if os.path.exists(path):
os.remove(path)
def genDetRecTrainVal(args):
detAbsTrainRootPath = isCreateOrDeleteFolder(args.detRootPath, "train")
detAbsValRootPath = isCreateOrDeleteFolder(args.detRootPath, "val")
detAbsTestRootPath = isCreateOrDeleteFolder(args.detRootPath, "test")
recAbsTrainRootPath = isCreateOrDeleteFolder(args.recRootPath, "train")
recAbsValRootPath = isCreateOrDeleteFolder(args.recRootPath, "val")
recAbsTestRootPath = isCreateOrDeleteFolder(args.recRootPath, "test")
removeFile(os.path.join(args.detRootPath, "train.txt"))
removeFile(os.path.join(args.detRootPath, "val.txt"))
removeFile(os.path.join(args.detRootPath, "test.txt"))
removeFile(os.path.join(args.recRootPath, "train.txt"))
removeFile(os.path.join(args.recRootPath, "val.txt"))
removeFile(os.path.join(args.recRootPath, "test.txt"))
detTrainTxt = open(os.path.join(args.detRootPath, "train.txt"), "a", encoding="UTF-8")
detValTxt = open(os.path.join(args.detRootPath, "val.txt"), "a", encoding="UTF-8")
detTestTxt = open(os.path.join(args.detRootPath, "test.txt"), "a", encoding="UTF-8")
recTrainTxt = open(os.path.join(args.recRootPath, "train.txt"), "a", encoding="UTF-8")
recValTxt = open(os.path.join(args.recRootPath, "val.txt"), "a", encoding="UTF-8")
recTestTxt = open(os.path.join(args.recRootPath, "test.txt"), "a", encoding="UTF-8")
for root, dirs, files in os.walk(args.labelRootPath):
for dir in dirs:
splitTrainVal(root, dir, detAbsTrainRootPath, detAbsValRootPath, detAbsTestRootPath, detTrainTxt, detValTxt,
detTestTxt, "det")
splitTrainVal(root, dir, recAbsTrainRootPath, recAbsValRootPath, recAbsTestRootPath, recTrainTxt, recValTxt,
recTestTxt, "rec")
break
if __name__ == "__main__":
# 功能描述:分别划分检测和识别的训练集、验证集、测试集
# 说明:可以根据自己的路径和需求调整参数,图像数据往往多人合作分批标注,每一批图像数据放在一个文件夹内用PPOCRLabel进行标注,
# 如此会有多个标注好的图像文件夹汇总并划分训练集、验证集、测试集的需求
parser = argparse.ArgumentParser()
parser.add_argument(
"--trainValTestRatio",
type=str,
default="6:2:2",
help="ratio of trainset:valset:testset")
parser.add_argument(
"--labelRootPath",
type=str,
default="../train_data/label",
help="path to the dataset marked by ppocrlabel, E.g, dataset folder named 1,2,3..."
)
parser.add_argument(
"--detRootPath",
type=str,
default="../train_data/det",
help="the path where the divided detection dataset is placed")
parser.add_argument(
"--recRootPath",
type=str,
default="../train_data/rec",
help="the path where the divided recognition dataset is placed"
)
parser.add_argument(
"--detLabelFileName",
type=str,
default="Label.txt",
help="the name of the detection annotation file")
parser.add_argument(
"--recLabelFileName",
type=str,
default="rec_gt.txt",
help="the name of the recognition annotation file"
)
parser.add_argument(
"--recImageDirName",
type=str,
default="crop_img",
help="the name of the folder where the cropped recognition dataset is located"
)
args = parser.parse_args()
genDetRecTrainVal(args)
此差异已折叠。
......@@ -19,6 +19,9 @@ import sys
import locale
from libs.ustr import ustr
__dir__ = os.path.dirname(os.path.abspath(__file__)) # 获取本程序文件路径
__dirpath__ = os.path.abspath(os.path.join(__dir__, '../resources/strings'))
try:
from PyQt5.QtCore import *
except ImportError:
......@@ -57,7 +60,7 @@ class StringBundle:
def __createLookupFallbackList(self, localeStr):
resultPaths = []
basePath = ":/strings"
basePath = "\strings" if os.name == 'nt' else ":/strings"
resultPaths.append(basePath)
if localeStr is not None:
# Don't follow standard BCP47. Simple fallback
......@@ -65,6 +68,7 @@ class StringBundle:
for tag in tags:
lastPath = resultPaths[-1]
resultPaths.append(lastPath + '-' + tag)
resultPaths[-1] = __dirpath__ + resultPaths[-1] + ".properties"
return resultPaths
......
......@@ -17,6 +17,10 @@ import re
import sys
import cv2
import numpy as np
import os
__dir__ = os.path.dirname(os.path.abspath(__file__)) # 获取本程序文件路径
__iconpath__ = os.path.abspath(os.path.join(__dir__, '../resources/icons'))
try:
from PyQt5.QtGui import *
......@@ -29,9 +33,9 @@ except ImportError:
def newIcon(icon, iconSize=None):
if iconSize is not None:
return QIcon(QIcon(':/' + icon).pixmap(iconSize,iconSize))
return QIcon(QIcon(__iconpath__ + "/" + icon + ".png").pixmap(iconSize,iconSize))
else:
return QIcon(':/' + icon)
return QIcon(__iconpath__ + "/" + icon + ".png")
def newButton(text, icon=None, slot=None):
......
pyqt5
paddleocr
\ No newline at end of file
......@@ -35,7 +35,7 @@
<file alias="prev">resources/icons/prev.png</file>
<file alias="resetall">resources/icons/resetall.png</file>
<file alias="verify">resources/icons/verify.png</file>
<file alias="strings">resources/strings/strings.properties</file>
<file alias="strings">resources/strings/strings-en.properties</file>
<file alias="strings-zh-CN">resources/strings/strings-zh-CN.properties</file>
</qresource>
</RCC>
# Copyright (c) <2015-Present> Tzutalin
# Copyright (C) 2013 MIT, Computer Science and Artificial Intelligence Laboratory. Bryan Russell, Antonio Torralba,
# William T. Freeman. Permission is hereby granted, free of charge, to any person obtaining a copy of this software and
# associated documentation files (the "Software"), to deal in the Software without restriction, including without
# limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the
# Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
# The above copyright notice and this permission notice shall be included in all copies or substantial portions of
# the Software. THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT
# NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT
# SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF
# CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
# THE SOFTWARE.
#!/usr/bin/env python
# -*- coding: utf-8 -*-
from setuptools import setup, find_packages, Command
from sys import platform as _platform
from shutil import rmtree
import sys
import os
here = os.path.abspath(os.path.dirname(__file__))
NAME = 'labelImg'
REQUIRES_PYTHON = '>=3.0.0'
REQUIRED_DEP = ['pyqt5', 'lxml']
about = {}
with open(os.path.join(here, 'libs', '__init__.py')) as f:
exec(f.read(), about)
with open('README.rst') as readme_file:
readme = readme_file.read()
with open('HISTORY.rst') as history_file:
history = history_file.read()
# OS specific settings
SET_REQUIRES = []
if _platform == "linux" or _platform == "linux2":
# linux
print('linux')
elif _platform == "darwin":
# MAC OS X
SET_REQUIRES.append('py2app')
required_packages = find_packages()
required_packages.append('labelImg')
APP = [NAME + '.py']
OPTIONS = {
'argv_emulation': True,
'iconfile': 'resources/icons/app.icns'
}
class UploadCommand(Command):
"""Support setup.py upload."""
description=readme + '\n\n' + history,
user_options = []
@staticmethod
def status(s):
"""Prints things in bold."""
print('\033[1m{0}\033[0m'.format(s))
def initialize_options(self):
pass
def finalize_options(self):
pass
def run(self):
try:
self.status('Removing previous builds…')
rmtree(os.path.join(here, 'dist'))
except OSError:
self.status('Fail to remove previous builds..')
pass
self.status('Building Source and Wheel (universal) distribution…')
os.system(
'{0} setup.py sdist bdist_wheel --universal'.format(sys.executable))
self.status('Uploading the package to PyPI via Twine…')
os.system('twine upload dist/*')
self.status('Pushing git tags…')
os.system('git tag -d v{0}'.format(about['__version__']))
os.system('git tag v{0}'.format(about['__version__']))
# os.system('git push --tags')
sys.exit()
# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from setuptools import setup
from io import open
with open('requirements.txt', encoding="utf-8-sig") as f:
requirements = f.readlines()
requirements.append('tqdm')
def readme():
with open('README.md', encoding="utf-8-sig") as f:
README = f.read()
return README
setup(
app=APP,
name=NAME,
version=about['__version__'],
description="LabelImg is a graphical image annotation tool and label object bounding boxes in images",
long_description=readme + '\n\n' + history,
author="TzuTa Lin",
author_email='tzu.ta.lin@gmail.com',
url='https://github.com/tzutalin/labelImg',
python_requires=REQUIRES_PYTHON,
package_dir={'labelImg': '.'},
packages=required_packages,
entry_points={
'console_scripts': [
'labelImg=labelImg.labelImg:main'
]
},
name='PPOCRLabel',
packages=['PPOCRLabel'],
package_data = {'PPOCRLabel': ['libs/*','resources/strings/*','resources/icons/*']},
package_dir={'PPOCRLabel': ''},
include_package_data=True,
install_requires=REQUIRED_DEP,
license="MIT license",
zip_safe=False,
keywords='labelImg labelTool development annotation deeplearning',
entry_points={"console_scripts": ["PPOCRLabel= PPOCRLabel.PPOCRLabel:main"]},
version='1.0.0',
install_requires=requirements,
license='Apache License 2.0',
description='PPOCRLabel is a semi-automatic graphic annotation tool suitable for OCR field, with built-in PPOCR model to automatically detect and re-recognize data. It is written in python3 and pyqt5, supporting rectangular box annotation and four-point annotation modes. Annotations can be directly used for the training of PPOCR detection and recognition models',
long_description=readme(),
long_description_content_type='text/markdown',
url='https://github.com/PaddlePaddle/PaddleOCR',
download_url='https://github.com/PaddlePaddle/PaddleOCR.git',
keywords=[
'ocr textdetection textrecognition paddleocr crnn east star-net rosetta ocrlite db chineseocr chinesetextdetection chinesetextrecognition'
],
classifiers=[
'Development Status :: 5 - Production/Stable',
'Intended Audience :: Developers',
'License :: OSI Approved :: MIT License',
'Intended Audience :: Developers', 'Operating System :: OS Independent',
'Natural Language :: English',
'Programming Language :: Python :: 3',
'Programming Language :: Python :: 3.3',
'Programming Language :: Python :: 3.4',
'Programming Language :: Python :: 3.5',
'Programming Language :: Python :: 3.6',
'Programming Language :: Python :: 3.7',
],
package_data={'data/predefined_classes.txt': ['data/predefined_classes.txt']},
options={'py2app': OPTIONS},
setup_requires=SET_REQUIRES,
# $ setup.py publish support.
cmdclass={
'upload': UploadCommand,
}
)
'Programming Language :: Python :: 3.7', 'Topic :: Utilities'
], )
\ No newline at end of file
......@@ -25,7 +25,7 @@ PaddleOCR aims to create multilingual, awesome, leading, and practical OCR tools
**Recent updates**
- 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).
- PaddleOCR R&D team would like to share the key points of PP-OCRv2, at 20:15 pm on September 8th, [Course Address](https://aistudio.baidu.com/aistudio/education/group/info/6758).
- 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.
......@@ -86,9 +86,9 @@ Mobile DEMO experience (based on EasyEdge and Paddle-Lite, supports iOS and Andr
| Model introduction | Model name | Recommended scene | Detection model | Direction classifier | Recognition model |
| ------------------------------------------------------------ | ---------------------------- | ----------------- | ------------------------------------------------------------ | ------------------------------------------------------------ | ------------------------------------------------------------ |
| 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) |
| 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) / [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) / [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) / [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) / [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) / [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) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_train.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) / [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) / [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) / [trained model](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_rec_train.tar) |
For more model downloads (including multiple languages), please refer to [PP-OCR series model downloads](./doc/doc_en/models_list_en.md).
......@@ -102,7 +102,6 @@ For a new language request, please refer to [Guideline for new language_requests
- 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)
......@@ -119,7 +118,7 @@ For a new language request, please refer to [Guideline for new language_requests
- [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)
- [PGNet Algorithm](./doc/doc_en/pgnet_en.md)
- [Python Inference](./doc/doc_en/inference_en.md)
- Data Annotation and Synthesis
- [Semi-automatic Annotation Tool: PPOCRLabel](./PPOCRLabel/README.md)
......
......@@ -24,7 +24,7 @@ PaddleOCR旨在打造一套丰富、领先、且实用的OCR工具库,助力
**近期更新**
- 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)
- 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,[课程回放](https://aistudio.baidu.com/aistudio/education/group/info/6758)
- 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。
......@@ -81,9 +81,9 @@ PaddleOCR旨在打造一套丰富、领先、且实用的OCR工具库,助力
| 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/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) / [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) |
| 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/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) / [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) / [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) / [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) / [训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_rec_train.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) / [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) / [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) / [训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_server_v2.0_rec_train.tar) |
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)
......@@ -94,7 +94,6 @@ PaddleOCR旨在打造一套丰富、领先、且实用的OCR工具库,助力
- 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)
......@@ -109,15 +108,16 @@ PaddleOCR旨在打造一套丰富、领先、且实用的OCR工具库,助力
- [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)
- 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)
- [基于Python脚本预测引擎推理](./doc/doc_ch/inference.md)
- [使用PaddleOCR架构添加新算法](./doc/doc_ch/add_new_algorithm.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)
- [基于Python脚本预测引擎推理](./doc/doc_ch/inference.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)
......
# PaddleOCR DB/EAST 算法训练benchmark测试
# PaddleOCR DB/EAST/PSE 算法训练benchmark测试
PaddleOCR/benchmark目录下的文件用于获取并分析训练日志。
训练采用icdar2015数据集,包括1000张训练图像和500张测试图像。模型配置采用resnet18_vd作为backbone,分别训练batch_size=8和batch_size=16的情况。
......@@ -18,7 +18,7 @@ run_det.sh 执行方式如下:
```
# cd PaddleOCR/
bash benchmark/run_det.sh
bash benchmark/run_det.sh
```
以DB为例,将得到四个日志文件,如下:
......@@ -28,7 +28,3 @@ det_res18_db_v2.0_sp_bs8_fp32_1
det_res18_db_v2.0_mp_bs16_fp32_1
det_res18_db_v2.0_mp_bs8_fp32_1
```
......@@ -5,28 +5,36 @@ set -xe
function _set_params(){
run_mode=${1:-"sp"} # 单卡sp|多卡mp
batch_size=${2:-"64"}
fp_item=${3:-"fp32"} # fp32|fp16
max_iter=${4:-"500"} # 可选,如果需要修改代码提前中断
model_name=${5:-"model_name"}
fp_item=${3:-"fp32"} # fp32|fp16
max_epoch=${4:-"10"} # 可选,如果需要修改代码提前中断
model_item=${5:-"model_item"}
run_log_path=${TRAIN_LOG_DIR:-$(pwd)} # TRAIN_LOG_DIR 后续QA设置该参数
# 日志解析所需参数
base_batch_size=${batch_size}
mission_name="OCR"
direction_id="0"
ips_unit="images/sec"
skip_steps=2 # 解析日志,有些模型前几个step耗时长,需要跳过 (必填)
keyword="ips:" # 解析日志,筛选出数据所在行的关键字 (必填)
index="1"
model_name=${model_item}_${run_mode}_bs${batch_size}_${fp_item} # model_item 用于yml文件名匹配,model_name 用于数据入库前端展示
# 以下不用修改
device=${CUDA_VISIBLE_DEVICES//,/ }
arr=(${device})
num_gpu_devices=${#arr[*]}
log_file=${run_log_path}/${model_name}_${run_mode}_bs${batch_size}_${fp_item}_${num_gpu_devices}
log_file=${run_log_path}/${model_item}_${run_mode}_bs${batch_size}_${fp_item}_${num_gpu_devices}
}
function _train(){
echo "Train on ${num_gpu_devices} GPUs"
echo "current CUDA_VISIBLE_DEVICES=$CUDA_VISIBLE_DEVICES, gpus=$num_gpu_devices, batch_size=$batch_size"
train_cmd="-c configs/det/${model_name}.yml -o Train.loader.batch_size_per_card=${batch_size} Global.epoch_num=${max_iter} "
train_cmd="-c configs/det/${model_item}.yml -o Train.loader.batch_size_per_card=${batch_size} Global.epoch_num=${max_epoch} Global.eval_batch_step=[0,20000] Global.print_batch_step=2"
case ${run_mode} in
sp)
train_cmd="python3.7 tools/train.py "${train_cmd}""
train_cmd="python tools/train.py "${train_cmd}""
;;
mp)
train_cmd="python3.7 -m paddle.distributed.launch --log_dir=./mylog --gpus=$CUDA_VISIBLE_DEVICES tools/train.py ${train_cmd}"
train_cmd="python -m paddle.distributed.launch --log_dir=./mylog --gpus=$CUDA_VISIBLE_DEVICES tools/train.py ${train_cmd}"
;;
*) echo "choose run_mode(sp or mp)"; exit 1;
esac
......@@ -39,18 +47,14 @@ function _train(){
echo -e "${model_name}, SUCCESS"
export job_fail_flag=0
fi
kill -9 `ps -ef|grep 'python3.7'|awk '{print $2}'`
if [ $run_mode = "mp" -a -d mylog ]; then
rm ${log_file}
cp mylog/workerlog.0 ${log_file}
fi
# run log analysis
analysis_cmd="python3.7 benchmark/analysis.py --filename ${log_file} --mission_name ${model_name} --run_mode ${mode} --direction_id 0 --keyword 'ips:' --base_batch_size ${batch_szie} --skip_steps 1 --gpu_num ${num_gpu_devices} --index 1 --model_mode=-1 --ips_unit=samples/sec"
eval $analysis_cmd
}
source ${BENCHMARK_ROOT}/scripts/run_model.sh # 在该脚本中会对符合benchmark规范的log使用analysis.py 脚本进行性能数据解析;该脚本在连调时可从benchmark repo中下载https://github.com/PaddlePaddle/benchmark/blob/master/scripts/run_model.sh;如果不联调只想要产出训练log可以注掉本行,提交时需打开
_set_params $@
_train
#_train # 如果只想产出训练log,不解析,可取消注释
_run # 该函数在run_model.sh中,执行时会调用_train; 如果不联调只想要产出训练log可以注掉本行,提交时需打开
#!/bin/bash
# 提供可稳定复现性能的脚本,默认在标准docker环境内py37执行: paddlepaddle/paddle:latest-gpu-cuda10.1-cudnn7 paddle=2.1.2 py=37
# 执行目录: ./PaddleOCR
# 1 安装该模型需要的依赖 (如需开启优化策略请注明)
python3.7 -m pip install -r requirements.txt
python -m pip install -r requirements.txt
# 2 拷贝该模型需要数据、预训练模型
wget -c -p ./tain_data/ https://paddleocr.bj.bcebos.com/dygraph_v2.0/test/icdar2015.tar && cd train_data && tar xf icdar2015.tar && cd ../
wget -c -p ./pretrain_models/ https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet50_vd_pretrained.pdparams
wget -P ./train_data/ https://paddleocr.bj.bcebos.com/dygraph_v2.0/test/icdar2015.tar && cd train_data && tar xf icdar2015.tar && cd ../
wget -P ./pretrain_models/ https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet50_vd_pretrained.pdparams
wget -P ./pretrain_models/ https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet18_vd_pretrained.pdparams
wget -P ./pretrain_models/ https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/ResNet50_vd_ssld_pretrained.pdparams
# 3 批量运行(如不方便批量,1,2需放到单个模型中)
model_mode_list=(det_res18_db_v2.0 det_r50_vd_east)
model_mode_list=(det_res18_db_v2.0 det_r50_vd_east det_r50_vd_pse)
fp_item_list=(fp32)
bs_list=(8 16)
for model_mode in ${model_mode_list[@]}; do
......@@ -15,11 +18,13 @@ for model_mode in ${model_mode_list[@]}; do
for bs_item in ${bs_list[@]}; do
echo "index is speed, 1gpus, begin, ${model_name}"
run_mode=sp
CUDA_VISIBLE_DEVICES=0 bash benchmark/run_benchmark_det.sh ${run_mode} ${bs_item} ${fp_item} 10 ${model_mode} # (5min)
log_name=ocr_${model_mode}_${run_mode}_bs${bs_item}_${fp_item}
CUDA_VISIBLE_DEVICES=0 bash benchmark/run_benchmark_det.sh ${run_mode} ${bs_item} ${fp_item} 1 ${model_mode} | tee ${log_path}/${log_name}_speed_1gpus 2>&1 # (5min)
sleep 60
echo "index is speed, 8gpus, run_mode is multi_process, begin, ${model_name}"
run_mode=mp
CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7 bash benchmark/run_benchmark_det.sh ${run_mode} ${bs_item} ${fp_item} 10 ${model_mode}
log_name=ocr_${model_mode}_${run_mode}_bs${bs_item}_${fp_item}
CUDA_VISIBLE_DEVICES=0,1,2,3,4,5,6,7 bash benchmark/run_benchmark_det.sh ${run_mode} ${bs_item} ${fp_item} 2 ${model_mode} | tee ${log_path}/${log_name}_speed_8gpus8p 2>&1
sleep 60
done
done
......
......@@ -141,6 +141,7 @@ Train:
img_mode: BGR
channel_first: False
- DetLabelEncode: # Class handling label
- CopyPaste:
- IaaAugment:
augmenter_args:
- { 'type': Fliplr, 'args': { 'p': 0.5 } }
......
......@@ -68,8 +68,7 @@ Loss:
ohem_ratio: 3
- DistillationDBLoss:
weight: 1.0
model_name_list: ["Student", "Teacher"]
# key: maps
model_name_list: ["Student"]
name: DBLoss
balance_loss: true
main_loss_type: DiceLoss
......@@ -91,7 +90,7 @@ Optimizer:
PostProcess:
name: DistillationDBPostProcess
model_name: ["Student", "Student2"]
model_name: ["Student"]
key: head_out
thresh: 0.3
box_thresh: 0.6
......@@ -116,6 +115,7 @@ Train:
img_mode: BGR
channel_first: False
- DetLabelEncode: # Class handling label
- CopyPaste:
- IaaAugment:
augmenter_args:
- { 'type': Fliplr, 'args': { 'p': 0.5 } }
......
......@@ -118,6 +118,7 @@ Train:
img_mode: BGR
channel_first: False
- DetLabelEncode: # Class handling label
- CopyPaste:
- IaaAugment:
augmenter_args:
- { 'type': Fliplr, 'args': { 'p': 0.5 } }
......
......@@ -8,7 +8,7 @@ Global:
# evaluation is run every 5000 iterations after the 4000th iteration
eval_batch_step: [4000, 5000]
cal_metric_during_train: False
pretrained_model: ./pretrain_models/ResNet50_vd_ssld_pretrained/
pretrained_model: ./pretrain_models/ResNet50_vd_ssld_pretrained
checkpoints:
save_inference_dir:
use_visualdl: False
......@@ -106,4 +106,4 @@ Eval:
shuffle: False
drop_last: False
batch_size_per_card: 1 # must be 1
num_workers: 2
\ No newline at end of file
num_workers: 2
......@@ -8,7 +8,7 @@ Global:
# evaluation is run every 5000 iterations after the 4000th iteration
eval_batch_step: [4000, 5000]
cal_metric_during_train: False
pretrained_model: ./pretrain_models/ResNet50_vd_ssld_pretrained/
pretrained_model: ./pretrain_models/ResNet50_vd_ssld_pretrained
checkpoints:
save_inference_dir:
use_visualdl: False
......
......@@ -94,7 +94,7 @@ Eval:
label_file_list: [./train_data/total_text/test/test.txt]
transforms:
- DecodeImage: # load image
img_mode: RGB
img_mode: BGR
channel_first: False
- E2ELabelEncodeTest:
- E2EResizeForTest:
......
......@@ -14,7 +14,6 @@ Global:
use_visualdl: false
infer_img: doc/imgs_words/ch/word_1.jpg
character_dict_path: ppocr/utils/ppocr_keys_v1.txt
character_type: ch
max_text_length: 25
infer_mode: false
use_space_char: true
......
......@@ -14,7 +14,6 @@ Global:
use_visualdl: false
infer_img: doc/imgs_words/ch/word_1.jpg
character_dict_path: ppocr/utils/ppocr_keys_v1.txt
character_type: ch
max_text_length: 25
infer_mode: false
use_space_char: true
......
......@@ -14,7 +14,6 @@ Global:
use_visualdl: false
infer_img: doc/imgs_words/ch/word_1.jpg
character_dict_path: ppocr/utils/ppocr_keys_v1.txt
character_type: ch
max_text_length: 25
infer_mode: false
use_space_char: true
......@@ -63,8 +62,7 @@ Loss:
weight: 0.05
num_classes: 6625
feat_dim: 96
init_center: false
center_file_path: "./train_center.pkl"
center_file_path:
# you can also try to add ace loss on your own dataset
# - ACELoss:
# weight: 0.1
......
......@@ -15,7 +15,6 @@ Global:
infer_img: doc/imgs_words/ch/word_1.jpg
# for data or label process
character_dict_path: ppocr/utils/ppocr_keys_v1.txt
character_type: ch
max_text_length: 25
infer_mode: False
use_space_char: True
......
......@@ -15,7 +15,6 @@ Global:
infer_img: doc/imgs_words/ch/word_1.jpg
# for data or label process
character_dict_path: ppocr/utils/ppocr_keys_v1.txt
character_type: ch
max_text_length: 25
infer_mode: False
use_space_char: True
......
......@@ -15,7 +15,6 @@ Global:
use_visualdl: false
infer_img: null
character_dict_path: ppocr/utils/dict/arabic_dict.txt
character_type: arabic
max_text_length: 25
infer_mode: false
use_space_char: true
......
......@@ -15,7 +15,6 @@ Global:
use_visualdl: false
infer_img: null
character_dict_path: ppocr/utils/dict/cyrillic_dict.txt
character_type: cyrillic
max_text_length: 25
infer_mode: false
use_space_char: true
......
......@@ -15,7 +15,6 @@ Global:
use_visualdl: false
infer_img: null
character_dict_path: ppocr/utils/dict/devanagari_dict.txt
character_type: devanagari
max_text_length: 25
infer_mode: false
use_space_char: true
......
......@@ -16,7 +16,6 @@ Global:
infer_img:
# for data or label process
character_dict_path: ppocr/utils/en_dict.txt
character_type: EN
max_text_length: 25
infer_mode: False
use_space_char: True
......
......@@ -16,7 +16,6 @@ Global:
infer_img:
# for data or label process
character_dict_path: ppocr/utils/dict/french_dict.txt
character_type: french
max_text_length: 25
infer_mode: False
use_space_char: False
......
......@@ -16,7 +16,6 @@ Global:
infer_img:
# for data or label process
character_dict_path: ppocr/utils/dict/german_dict.txt
character_type: german
max_text_length: 25
infer_mode: False
use_space_char: False
......
......@@ -16,7 +16,6 @@ Global:
infer_img:
# for data or label process
character_dict_path: ppocr/utils/dict/japan_dict.txt
character_type: japan
max_text_length: 25
infer_mode: False
use_space_char: False
......
......@@ -16,7 +16,6 @@ Global:
infer_img:
# for data or label process
character_dict_path: ppocr/utils/dict/korean_dict.txt
character_type: korean
max_text_length: 25
infer_mode: False
use_space_char: False
......
......@@ -15,7 +15,6 @@ Global:
use_visualdl: false
infer_img: null
character_dict_path: ppocr/utils/dict/latin_dict.txt
character_type: latin
max_text_length: 25
infer_mode: false
use_space_char: true
......
......@@ -15,7 +15,6 @@ Global:
infer_img: doc/imgs_words_en/word_10.png
# for data or label process
character_dict_path: ppocr/utils/en_dict.txt
character_type: EN
max_text_length: 25
infer_mode: False
use_space_char: False
......
......@@ -14,11 +14,10 @@ Global:
use_visualdl: False
infer_img: doc/imgs_words_en/word_10.png
# for data or label process
character_dict_path:
character_type: EN_symbol
character_dict_path: ppocr/utils/EN_symbol_dict.txt
max_text_length: 25
infer_mode: False
use_space_char: True
use_space_char: False
save_res_path: ./output/rec/predicts_nrtr.txt
Optimizer:
......
......@@ -14,8 +14,7 @@ Global:
use_visualdl: False
infer_img: doc/imgs_words_en/word_10.png
# for data or label process
character_dict_path:
character_type: en
character_dict_path:
max_text_length: 25
infer_mode: False
use_space_char: False
......
......@@ -15,7 +15,6 @@ Global:
infer_img: doc/imgs_words_en/word_10.png
# for data or label process
character_dict_path:
character_type: en
max_text_length: 25
infer_mode: False
use_space_char: False
......
......@@ -14,8 +14,7 @@ Global:
use_visualdl: False
infer_img: doc/imgs_words/ch/word_1.jpg
# for data or label process
character_dict_path:
character_type: en
character_dict_path:
max_text_length: 25
infer_mode: False
use_space_char: False
......
......@@ -15,7 +15,6 @@ Global:
infer_img: doc/imgs_words_en/word_10.png
# for data or label process
character_dict_path:
character_type: en
max_text_length: 25
infer_mode: False
use_space_char: False
......
......@@ -15,7 +15,6 @@ Global:
infer_img:
# for data or label process
character_dict_path: ppocr/utils/dict90.txt
character_type: EN_symbol
max_text_length: 30
infer_mode: False
use_space_char: False
......
......@@ -14,8 +14,7 @@ Global:
use_visualdl: False
infer_img: doc/imgs_words_en/word_10.png
# for data or label process
character_dict_path:
character_type: en
character_dict_path:
max_text_length: 25
infer_mode: False
use_space_char: False
......
......@@ -15,7 +15,6 @@ Global:
infer_img: doc/imgs_words_en/word_10.png
# for data or label process
character_dict_path:
character_type: en
max_text_length: 25
infer_mode: False
use_space_char: False
......
......@@ -14,8 +14,7 @@ Global:
use_visualdl: False
infer_img: doc/imgs_words/ch/word_1.jpg
# for data or label process
character_dict_path:
character_type: en
character_dict_path:
max_text_length: 25
infer_mode: False
use_space_char: False
......
......@@ -14,8 +14,7 @@ Global:
use_visualdl: False
infer_img: doc/imgs_words_en/word_10.png
# for data or label process
character_dict_path:
character_type: en
character_dict_path:
max_text_length: 25
infer_mode: False
use_space_char: False
......
......@@ -14,8 +14,7 @@ Global:
use_visualdl: False
infer_img: doc/imgs_words/ch/word_1.jpg
# for data or label process
character_dict_path:
character_type: en
character_dict_path:
max_text_length: 25
num_heads: 8
infer_mode: False
......
......@@ -14,8 +14,7 @@ Global:
use_visualdl: False
infer_img: doc/imgs_words_en/word_10.png
# for data or label process
character_dict_path:
character_type: EN_symbol
character_dict_path: ppocr/utils/EN_symbol_dict.txt
max_text_length: 100
infer_mode: False
use_space_char: False
......
Global:
use_gpu: true
epoch_num: 50
epoch_num: 400
log_smooth_window: 20
print_batch_step: 5
save_model_dir: ./output/table_mv3/
save_epoch_step: 5
save_epoch_step: 3
# evaluation is run every 400 iterations after the 0th iteration
eval_batch_step: [0, 400]
cal_metric_during_train: True
pretrained_model:
pretrained_model:
checkpoints:
save_inference_dir:
use_visualdl: False
infer_img: doc/imgs_words/ch/word_1.jpg
infer_img: doc/table/table.jpg
# for data or label process
character_dict_path: ppocr/utils/dict/table_structure_dict.txt
character_type: en
max_text_length: 100
max_elem_length: 500
max_elem_length: 800
max_cell_num: 500
infer_mode: False
process_total_num: 0
process_cut_num: 0
Optimizer:
name: Adam
beta1: 0.9
......@@ -41,13 +40,15 @@ Architecture:
Backbone:
name: MobileNetV3
scale: 1.0
model_name: small
disable_se: True
model_name: large
Head:
name: TableAttentionHead
hidden_size: 256
l2_decay: 0.00001
loc_type: 2
max_text_length: 100
max_elem_length: 800
max_cell_num: 500
Loss:
name: TableAttentionLoss
......
......@@ -44,7 +44,8 @@ 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 std::string &precision) {
const bool &use_tensorrt, const std::string &precision,
const int &rec_batch_num) {
this->use_gpu_ = use_gpu;
this->gpu_id_ = gpu_id;
this->gpu_mem_ = gpu_mem;
......@@ -52,6 +53,7 @@ public:
this->use_mkldnn_ = use_mkldnn;
this->use_tensorrt_ = use_tensorrt;
this->precision_ = precision;
this->rec_batch_num_ = rec_batch_num;
this->label_list_ = Utility::ReadDict(label_path);
this->label_list_.insert(this->label_list_.begin(),
......@@ -64,7 +66,7 @@ public:
// Load Paddle inference model
void LoadModel(const std::string &model_dir);
void Run(cv::Mat &img, std::vector<double> *times);
void Run(std::vector<cv::Mat> img_list, std::vector<double> *times);
private:
std::shared_ptr<Predictor> predictor_;
......@@ -82,10 +84,12 @@ private:
bool is_scale_ = true;
bool use_tensorrt_ = false;
std::string precision_ = "fp32";
int rec_batch_num_ = 6;
// pre-process
CrnnResizeImg resize_op_;
Normalize normalize_op_;
Permute permute_op_;
PermuteBatch permute_op_;
// post-process
PostProcessor post_processor_;
......
......@@ -44,6 +44,11 @@ public:
virtual void Run(const cv::Mat *im, float *data);
};
class PermuteBatch {
public:
virtual void Run(const std::vector<cv::Mat> imgs, float *data);
};
class ResizeImgType0 {
public:
virtual void Run(const cv::Mat &img, cv::Mat &resize_img, int max_size_len,
......
......@@ -50,6 +50,9 @@ public:
static cv::Mat GetRotateCropImage(const cv::Mat &srcimage,
std::vector<std::vector<int>> box);
static std::vector<int> argsort(const std::vector<float>& array);
};
} // namespace PaddleOCR
\ No newline at end of file
......@@ -34,10 +34,10 @@ PaddleOCR模型部署。
* 首先需要从opencv官网上下载在Linux环境下源码编译的包,以opencv3.4.7为例,下载命令如下。
```
```bash
cd deploy/cpp_infer
wget https://github.com/opencv/opencv/archive/3.4.7.tar.gz
tar -xf 3.4.7.tar.gz
wget https://paddleocr.bj.bcebos.com/libs/opencv/opencv-3.4.7.tar.gz
tar -xf opencv-3.4.7.tar.gz
```
最终可以在当前目录下看到`opencv-3.4.7/`的文件夹。
......@@ -45,12 +45,13 @@ tar -xf 3.4.7.tar.gz
* 编译opencv,设置opencv源码路径(`root_path`)以及安装路径(`install_path`)。进入opencv源码路径下,按照下面的方式进行编译。
```shell
root_path=your_opencv_root_path
root_path="your_opencv_root_path"
install_path=${root_path}/opencv3
build_dir=${root_path}/build
rm -rf build
mkdir build
cd build
rm -rf ${build_dir}
mkdir ${build_dir}
cd ${build_dir}
cmake .. \
-DCMAKE_INSTALL_PREFIX=${install_path} \
......@@ -74,6 +75,11 @@ make -j
make install
```
也可以直接修改`tools/build_opencv.sh`的内容,然后直接运行下面的命令进行编译。
```shell
sh tools/build_opencv.sh
```
其中`root_path`为下载的opencv源码路径,`install_path`为opencv的安装路径,`make install`完成之后,会在该文件夹下生成opencv头文件和库文件,用于后面的OCR代码编译。
......@@ -233,12 +239,12 @@ CUDNN_LIB_DIR=/your_cudnn_lib_dir
--image_dir=../../doc/imgs/12.jpg
```
更多参数如下:
更多支持的可调节参数解释如下:
- 通用参数
|参数名称|类型|默认参数|意义|
| --- | --- | --- | --- |
| :---: | :---: | :---: | :---: |
|use_gpu|bool|false|是否使用GPU|
|gpu_id|int|0|GPU id,使用GPU时有效|
|gpu_mem|int|4000|申请的GPU内存|
......@@ -248,7 +254,7 @@ CUDNN_LIB_DIR=/your_cudnn_lib_dir
- 检测模型相关
|参数名称|类型|默认参数|意义|
| --- | --- | --- | --- |
| :---: | :---: | :---: | :---: |
|det_model_dir|string|-|检测模型inference model地址|
|max_side_len|int|960|输入图像长宽大于960时,等比例缩放图像,使得图像最长边为960|
|det_db_thresh|float|0.3|用于过滤DB预测的二值化图像,设置为0.-0.3对结果影响不明显|
......@@ -260,7 +266,7 @@ CUDNN_LIB_DIR=/your_cudnn_lib_dir
- 方向分类器相关
|参数名称|类型|默认参数|意义|
| --- | --- | --- | --- |
| :---: | :---: | :---: | :---: |
|use_angle_cls|bool|false|是否使用方向分类器|
|cls_model_dir|string|-|方向分类器inference model地址|
|cls_thresh|float|0.9|方向分类器的得分阈值|
......@@ -268,7 +274,7 @@ CUDNN_LIB_DIR=/your_cudnn_lib_dir
- 识别模型相关
|参数名称|类型|默认参数|意义|
| --- | --- | --- | --- |
| :---: | :---: | :---: | :---: |
|rec_model_dir|string|-|识别模型inference model地址|
|char_list_file|string|../../ppocr/utils/ppocr_keys_v1.txt|字典文件|
......
......@@ -17,10 +17,10 @@ PaddleOCR model deployment.
* First of all, you need to download the source code compiled package in the Linux environment from the opencv official website. Taking opencv3.4.7 as an example, the download command is as follows.
```
```bash
cd deploy/cpp_infer
wget https://github.com/opencv/opencv/archive/3.4.7.tar.gz
tar -xf 3.4.7.tar.gz
wget https://paddleocr.bj.bcebos.com/libs/opencv/opencv-3.4.7.tar.gz
tar -xf opencv-3.4.7.tar.gz
```
Finally, you can see the folder of `opencv-3.4.7/` in the current directory.
......
......@@ -61,7 +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_int32(rec_batch_num, 6, "rec_batch_num.");
DEFINE_string(char_list_file, "../../ppocr/utils/ppocr_keys_v1.txt", "Path of dictionary.");
......@@ -146,8 +146,9 @@ int main_rec(std::vector<cv::String> cv_all_img_names) {
CRNNRecognizer rec(FLAGS_rec_model_dir, FLAGS_use_gpu, FLAGS_gpu_id,
FLAGS_gpu_mem, FLAGS_cpu_threads,
FLAGS_enable_mkldnn, char_list_file,
FLAGS_use_tensorrt, FLAGS_precision);
FLAGS_use_tensorrt, FLAGS_precision, FLAGS_rec_batch_num);
std::vector<cv::Mat> img_list;
for (int i = 0; i < cv_all_img_names.size(); ++i) {
LOG(INFO) << "The predict img: " << cv_all_img_names[i];
......@@ -156,22 +157,21 @@ int main_rec(std::vector<cv::String> cv_all_img_names) {
std::cerr << "[ERROR] image read failed! image path: " << cv_all_img_names[i] << endl;
exit(1);
}
std::vector<double> rec_times;
rec.Run(srcimg, &rec_times);
time_info[0] += rec_times[0];
time_info[1] += rec_times[1];
time_info[2] += rec_times[2];
img_list.push_back(srcimg);
}
std::vector<double> rec_times;
rec.Run(img_list, &rec_times);
time_info[0] += rec_times[0];
time_info[1] += rec_times[1];
time_info[2] += rec_times[2];
if (FLAGS_benchmark) {
AutoLogger autolog("ocr_rec",
FLAGS_use_gpu,
FLAGS_use_tensorrt,
FLAGS_enable_mkldnn,
FLAGS_cpu_threads,
1,
FLAGS_rec_batch_num,
"dynamic",
FLAGS_precision,
time_info,
......@@ -209,7 +209,7 @@ int main_system(std::vector<cv::String> cv_all_img_names) {
CRNNRecognizer rec(FLAGS_rec_model_dir, FLAGS_use_gpu, FLAGS_gpu_id,
FLAGS_gpu_mem, FLAGS_cpu_threads,
FLAGS_enable_mkldnn, char_list_file,
FLAGS_use_tensorrt, FLAGS_precision);
FLAGS_use_tensorrt, FLAGS_precision, FLAGS_rec_batch_num);
for (int i = 0; i < cv_all_img_names.size(); ++i) {
LOG(INFO) << "The predict img: " << cv_all_img_names[i];
......@@ -228,19 +228,22 @@ int main_system(std::vector<cv::String> cv_all_img_names) {
time_info_det[1] += det_times[1];
time_info_det[2] += det_times[2];
cv::Mat crop_img;
std::vector<cv::Mat> img_list;
for (int j = 0; j < boxes.size(); j++) {
crop_img = Utility::GetRotateCropImage(srcimg, boxes[j]);
if (cls != nullptr) {
crop_img = cls->Run(crop_img);
}
rec.Run(crop_img, &rec_times);
time_info_rec[0] += rec_times[0];
time_info_rec[1] += rec_times[1];
time_info_rec[2] += rec_times[2];
cv::Mat crop_img;
crop_img = Utility::GetRotateCropImage(srcimg, boxes[j]);
if (cls != nullptr) {
crop_img = cls->Run(crop_img);
}
img_list.push_back(crop_img);
}
rec.Run(img_list, &rec_times);
time_info_rec[0] += rec_times[0];
time_info_rec[1] += rec_times[1];
time_info_rec[2] += rec_times[2];
}
if (FLAGS_benchmark) {
AutoLogger autolog_det("ocr_det",
FLAGS_use_gpu,
......@@ -257,7 +260,7 @@ int main_system(std::vector<cv::String> cv_all_img_names) {
FLAGS_use_tensorrt,
FLAGS_enable_mkldnn,
FLAGS_cpu_threads,
1,
FLAGS_rec_batch_num,
"dynamic",
FLAGS_precision,
time_info_rec,
......
......@@ -15,83 +15,108 @@
#include <include/ocr_rec.h>
namespace PaddleOCR {
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_,
this->is_scale_);
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> predict_batch;
auto output_names = this->predictor_->GetOutputNames();
auto output_t = this->predictor_->GetOutputHandle(output_names[0]);
auto predict_shape = output_t->shape();
int out_num = std::accumulate(predict_shape.begin(), predict_shape.end(), 1,
void CRNNRecognizer::Run(std::vector<cv::Mat> img_list, std::vector<double> *times) {
std::chrono::duration<float> preprocess_diff = std::chrono::steady_clock::now() - std::chrono::steady_clock::now();
std::chrono::duration<float> inference_diff = std::chrono::steady_clock::now() - std::chrono::steady_clock::now();
std::chrono::duration<float> postprocess_diff = std::chrono::steady_clock::now() - std::chrono::steady_clock::now();
int img_num = img_list.size();
std::vector<float> width_list;
for (int i = 0; i < img_num; i++) {
width_list.push_back(float(img_list[i].cols) / img_list[i].rows);
}
std::vector<int> indices = Utility::argsort(width_list);
for (int beg_img_no = 0; beg_img_no < img_num; beg_img_no += this->rec_batch_num_) {
auto preprocess_start = std::chrono::steady_clock::now();
int end_img_no = min(img_num, beg_img_no + this->rec_batch_num_);
float max_wh_ratio = 0;
for (int ino = beg_img_no; ino < end_img_no; ino ++) {
int h = img_list[indices[ino]].rows;
int w = img_list[indices[ino]].cols;
float wh_ratio = w * 1.0 / h;
max_wh_ratio = max(max_wh_ratio, wh_ratio);
}
std::vector<cv::Mat> norm_img_batch;
for (int ino = beg_img_no; ino < end_img_no; ino ++) {
cv::Mat srcimg;
img_list[indices[ino]].copyTo(srcimg);
cv::Mat resize_img;
this->resize_op_.Run(srcimg, resize_img, max_wh_ratio, this->use_tensorrt_);
this->normalize_op_.Run(&resize_img, this->mean_, this->scale_, this->is_scale_);
norm_img_batch.push_back(resize_img);
}
int batch_width = int(ceilf(32 * max_wh_ratio)) - 1;
std::vector<float> input(this->rec_batch_num_ * 3 * 32 * batch_width, 0.0f);
this->permute_op_.Run(norm_img_batch, input.data());
auto preprocess_end = std::chrono::steady_clock::now();
preprocess_diff += preprocess_end - preprocess_start;
// Inference.
auto input_names = this->predictor_->GetInputNames();
auto input_t = this->predictor_->GetInputHandle(input_names[0]);
input_t->Reshape({this->rec_batch_num_, 3, 32, batch_width});
auto inference_start = std::chrono::steady_clock::now();
input_t->CopyFromCpu(input.data());
this->predictor_->Run();
std::vector<float> predict_batch;
auto output_names = this->predictor_->GetOutputNames();
auto output_t = this->predictor_->GetOutputHandle(output_names[0]);
auto predict_shape = output_t->shape();
int out_num = std::accumulate(predict_shape.begin(), predict_shape.end(), 1,
std::multiplies<int>());
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;
float score = 0.f;
int count = 0;
float max_value = 0.0f;
for (int n = 0; n < predict_shape[1]; n++) {
argmax_idx =
int(Utility::argmax(&predict_batch[n * predict_shape[2]],
&predict_batch[(n + 1) * predict_shape[2]]));
max_value =
float(*std::max_element(&predict_batch[n * predict_shape[2]],
&predict_batch[(n + 1) * predict_shape[2]]));
if (argmax_idx > 0 && (!(n > 0 && argmax_idx == last_index))) {
score += max_value;
count += 1;
str_res.push_back(label_list_[argmax_idx]);
predict_batch.resize(out_num);
output_t->CopyToCpu(predict_batch.data());
auto inference_end = std::chrono::steady_clock::now();
inference_diff += inference_end - inference_start;
// ctc decode
auto postprocess_start = std::chrono::steady_clock::now();
for (int m = 0; m < predict_shape[0]; m++) {
std::vector<std::string> str_res;
int argmax_idx;
int last_index = 0;
float score = 0.f;
int count = 0;
float max_value = 0.0f;
for (int n = 0; n < predict_shape[1]; n++) {
argmax_idx =
int(Utility::argmax(&predict_batch[(m * predict_shape[1] + n) * predict_shape[2]],
&predict_batch[(m * predict_shape[1] + n + 1) * predict_shape[2]]));
max_value =
float(*std::max_element(&predict_batch[(m * predict_shape[1] + n) * predict_shape[2]],
&predict_batch[(m * predict_shape[1] + n + 1) * predict_shape[2]]));
if (argmax_idx > 0 && (!(n > 0 && argmax_idx == last_index))) {
score += max_value;
count += 1;
str_res.push_back(label_list_[argmax_idx]);
}
last_index = argmax_idx;
}
score /= count;
if (isnan(score))
continue;
for (int i = 0; i < str_res.size(); i++) {
std::cout << str_res[i];
}
std::cout << "\tscore: " << score << std::endl;
}
auto postprocess_end = std::chrono::steady_clock::now();
postprocess_diff += postprocess_end - postprocess_start;
}
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));
times->push_back(double(preprocess_diff.count() * 1000));
times->push_back(double(inference_diff.count() * 1000));
times->push_back(double(postprocess_diff.count() * 1000));
}
void CRNNRecognizer::LoadModel(const std::string &model_dir) {
// AnalysisConfig config;
paddle_infer::Config config;
......
......@@ -40,6 +40,17 @@ void Permute::Run(const cv::Mat *im, float *data) {
}
}
void PermuteBatch::Run(const std::vector<cv::Mat> imgs, float *data) {
for (int j = 0; j < imgs.size(); j ++){
int rh = imgs[j].rows;
int rw = imgs[j].cols;
int rc = imgs[j].channels();
for (int i = 0; i < rc; ++i) {
cv::extractChannel(imgs[j], cv::Mat(rh, rw, CV_32FC1, data + (j * rc + i) * rh * rw), i);
}
}
}
void Normalize::Run(cv::Mat *im, const std::vector<float> &mean,
const std::vector<float> &scale, const bool is_scale) {
double e = 1.0;
......@@ -90,16 +101,17 @@ void CrnnResizeImg::Run(const cv::Mat &img, cv::Mat &resize_img, float wh_ratio,
imgC = rec_image_shape[0];
imgH = rec_image_shape[1];
imgW = rec_image_shape[2];
imgW = int(32 * wh_ratio);
float ratio = float(img.cols) / float(img.rows);
int resize_w, resize_h;
if (ceilf(imgH * ratio) > imgW)
resize_w = imgW;
else
resize_w = int(ceilf(imgH * ratio));
cv::resize(img, resize_img, cv::Size(resize_w, imgH), 0.f, 0.f,
cv::INTER_LINEAR);
cv::copyMakeBorder(resize_img, resize_img, 0, 0, 0,
......
......@@ -147,4 +147,17 @@ cv::Mat Utility::GetRotateCropImage(const cv::Mat &srcimage,
}
}
std::vector<int> Utility::argsort(const std::vector<float>& array)
{
const int array_len(array.size());
std::vector<int> array_index(array_len, 0);
for (int i = 0; i < array_len; ++i)
array_index[i] = i;
std::sort(array_index.begin(), array_index.end(),
[&array](int pos1, int pos2) {return (array[pos1] < array[pos2]); });
return array_index;
}
} // namespace PaddleOCR
\ No newline at end of file
root_path="/paddle/PaddleOCR/deploy/cpp_infer/opencv-3.4.7"
install_path=${root_path}/opencv3
build_dir=${root_path}/build
rm -rf ${build_dir}
mkdir ${build_dir}
cd ${build_dir}
cmake .. \
-DCMAKE_INSTALL_PREFIX=${install_path} \
-DCMAKE_BUILD_TYPE=Release \
-DBUILD_SHARED_LIBS=OFF \
-DWITH_IPP=OFF \
-DBUILD_IPP_IW=OFF \
-DWITH_LAPACK=OFF \
-DWITH_EIGEN=OFF \
-DCMAKE_INSTALL_LIBDIR=lib64 \
-DWITH_ZLIB=ON \
-DBUILD_ZLIB=ON \
-DWITH_JPEG=ON \
-DBUILD_JPEG=ON \
-DWITH_PNG=ON \
-DBUILD_PNG=ON \
-DWITH_TIFF=ON \
-DBUILD_TIFF=ON
make -j
make install
......@@ -12,12 +12,14 @@
// See the License for the specific language governing permissions and
// limitations under the License.
#include "paddle_api.h" // NOLINT
#include <chrono>
#include "paddle_api.h" // NOLINT
#include "paddle_place.h"
#include "cls_process.h"
#include "crnn_process.h"
#include "db_post_process.h"
#include "AutoLog/auto_log/lite_autolog.h"
using namespace paddle::lite_api; // NOLINT
using namespace std;
......@@ -27,7 +29,7 @@ void NeonMeanScale(const float *din, float *dout, int size,
const std::vector<float> mean,
const std::vector<float> scale) {
if (mean.size() != 3 || scale.size() != 3) {
std::cerr << "[ERROR] mean or scale size must equal to 3\n";
std::cerr << "[ERROR] mean or scale size must equal to 3" << std::endl;
exit(1);
}
float32x4_t vmean0 = vdupq_n_f32(mean[0]);
......@@ -159,7 +161,8 @@ void RunRecModel(std::vector<std::vector<std::vector<int>>> boxes, cv::Mat img,
std::vector<float> &rec_text_score,
std::vector<std::string> charactor_dict,
std::shared_ptr<PaddlePredictor> predictor_cls,
int use_direction_classify) {
int use_direction_classify,
std::vector<double> *times) {
std::vector<float> mean = {0.5f, 0.5f, 0.5f};
std::vector<float> scale = {1 / 0.5f, 1 / 0.5f, 1 / 0.5f};
......@@ -169,7 +172,10 @@ void RunRecModel(std::vector<std::vector<std::vector<int>>> boxes, cv::Mat img,
cv::Mat resize_img;
int index = 0;
std::vector<double> time_info = {0, 0, 0};
for (int i = boxes.size() - 1; i >= 0; i--) {
auto preprocess_start = std::chrono::steady_clock::now();
crop_img = GetRotateCropImage(srcimg, boxes[i]);
if (use_direction_classify >= 1) {
crop_img = RunClsModel(crop_img, predictor_cls);
......@@ -188,7 +194,9 @@ void RunRecModel(std::vector<std::vector<std::vector<int>>> boxes, cv::Mat img,
auto *data0 = input_tensor0->mutable_data<float>();
NeonMeanScale(dimg, data0, resize_img.rows * resize_img.cols, mean, scale);
auto preprocess_end = std::chrono::steady_clock::now();
//// Run CRNN predictor
auto inference_start = std::chrono::steady_clock::now();
predictor_crnn->Run();
// Get output and run postprocess
......@@ -196,8 +204,10 @@ void RunRecModel(std::vector<std::vector<std::vector<int>>> boxes, cv::Mat img,
std::move(predictor_crnn->GetOutput(0)));
auto *predict_batch = output_tensor0->data<float>();
auto predict_shape = output_tensor0->shape();
auto inference_end = std::chrono::steady_clock::now();
// ctc decode
auto postprocess_start = std::chrono::steady_clock::now();
std::string str_res;
int argmax_idx;
int last_index = 0;
......@@ -221,19 +231,33 @@ void RunRecModel(std::vector<std::vector<std::vector<int>>> boxes, cv::Mat img,
score /= count;
rec_text.push_back(str_res);
rec_text_score.push_back(score);
auto postprocess_end = std::chrono::steady_clock::now();
std::chrono::duration<float> preprocess_diff = preprocess_end - preprocess_start;
time_info[0] += double(preprocess_diff.count() * 1000);
std::chrono::duration<float> inference_diff = inference_end - inference_start;
time_info[1] += double(inference_diff.count() * 1000);
std::chrono::duration<float> postprocess_diff = postprocess_end - postprocess_start;
time_info[2] += double(postprocess_diff.count() * 1000);
}
times->push_back(time_info[0]);
times->push_back(time_info[1]);
times->push_back(time_info[2]);
}
std::vector<std::vector<std::vector<int>>>
RunDetModel(std::shared_ptr<PaddlePredictor> predictor, cv::Mat img,
std::map<std::string, double> Config) {
std::map<std::string, double> Config, std::vector<double> *times) {
// Read img
int max_side_len = int(Config["max_side_len"]);
int det_db_use_dilate = int(Config["det_db_use_dilate"]);
cv::Mat srcimg;
img.copyTo(srcimg);
auto preprocess_start = std::chrono::steady_clock::now();
std::vector<float> ratio_hw;
img = DetResizeImg(img, max_side_len, ratio_hw);
cv::Mat img_fp;
......@@ -248,8 +272,10 @@ RunDetModel(std::shared_ptr<PaddlePredictor> predictor, cv::Mat img,
std::vector<float> scale = {1 / 0.229f, 1 / 0.224f, 1 / 0.225f};
const float *dimg = reinterpret_cast<const float *>(img_fp.data);
NeonMeanScale(dimg, data0, img_fp.rows * img_fp.cols, mean, scale);
auto preprocess_end = std::chrono::steady_clock::now();
// Run predictor
auto inference_start = std::chrono::steady_clock::now();
predictor->Run();
// Get output and post process
......@@ -257,8 +283,10 @@ RunDetModel(std::shared_ptr<PaddlePredictor> predictor, cv::Mat img,
std::move(predictor->GetOutput(0)));
auto *outptr = output_tensor->data<float>();
auto shape_out = output_tensor->shape();
auto inference_end = std::chrono::steady_clock::now();
// Save output
auto postprocess_start = std::chrono::steady_clock::now();
float pred[shape_out[2] * shape_out[3]];
unsigned char cbuf[shape_out[2] * shape_out[3]];
......@@ -287,14 +315,23 @@ RunDetModel(std::shared_ptr<PaddlePredictor> predictor, cv::Mat img,
std::vector<std::vector<std::vector<int>>> filter_boxes =
FilterTagDetRes(boxes, ratio_hw[0], ratio_hw[1], srcimg);
auto postprocess_end = std::chrono::steady_clock::now();
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));
return filter_boxes;
}
std::shared_ptr<PaddlePredictor> loadModel(std::string model_file) {
std::shared_ptr<PaddlePredictor> loadModel(std::string model_file, int num_threads) {
MobileConfig config;
config.set_model_from_file(model_file);
config.set_threads(num_threads);
std::shared_ptr<PaddlePredictor> predictor =
CreatePaddlePredictor<MobileConfig>(config);
return predictor;
......@@ -354,60 +391,285 @@ std::map<std::string, double> LoadConfigTxt(std::string config_path) {
return dict;
}
int main(int argc, char **argv) {
if (argc < 5) {
std::cerr << "[ERROR] usage: " << argv[0]
<< " det_model_file cls_model_file rec_model_file image_path "
"charactor_dict\n";
void check_params(int argc, char **argv) {
if (argc<=1 || (strcmp(argv[1], "det")!=0 && strcmp(argv[1], "rec")!=0 && strcmp(argv[1], "system")!=0)) {
std::cerr << "Please choose one mode of [det, rec, system] !" << std::endl;
exit(1);
}
std::string det_model_file = argv[1];
std::string rec_model_file = argv[2];
std::string cls_model_file = argv[3];
std::string img_path = argv[4];
std::string dict_path = argv[5];
if (strcmp(argv[1], "det") == 0) {
if (argc < 9){
std::cerr << "[ERROR] usage:" << argv[0]
<< " det det_model runtime_device num_threads batchsize img_dir det_config lite_benchmark_value" << std::endl;
exit(1);
}
}
if (strcmp(argv[1], "rec") == 0) {
if (argc < 9){
std::cerr << "[ERROR] usage:" << argv[0]
<< " rec rec_model runtime_device num_threads batchsize img_dir key_txt lite_benchmark_value" << std::endl;
exit(1);
}
}
if (strcmp(argv[1], "system") == 0) {
if (argc < 12){
std::cerr << "[ERROR] usage:" << argv[0]
<< " system det_model rec_model clas_model runtime_device num_threads batchsize img_dir det_config key_txt lite_benchmark_value" << std::endl;
exit(1);
}
}
}
void system(char **argv){
std::string det_model_file = argv[2];
std::string rec_model_file = argv[3];
std::string cls_model_file = argv[4];
std::string runtime_device = argv[5];
std::string precision = argv[6];
std::string num_threads = argv[7];
std::string batchsize = argv[8];
std::string img_dir = argv[9];
std::string det_config_path = argv[10];
std::string dict_path = argv[11];
if (strcmp(argv[6], "FP32") != 0 && strcmp(argv[6], "INT8") != 0) {
std::cerr << "Only support FP32 or INT8." << std::endl;
exit(1);
}
std::vector<cv::String> cv_all_img_names;
cv::glob(img_dir, cv_all_img_names);
//// load config from txt file
auto Config = LoadConfigTxt("./config.txt");
auto Config = LoadConfigTxt(det_config_path);
int use_direction_classify = int(Config["use_direction_classify"]);
auto start = std::chrono::system_clock::now();
auto charactor_dict = ReadDict(dict_path);
charactor_dict.insert(charactor_dict.begin(), "#"); // blank char for ctc
charactor_dict.push_back(" ");
auto det_predictor = loadModel(det_model_file, std::stoi(num_threads));
auto rec_predictor = loadModel(rec_model_file, std::stoi(num_threads));
auto cls_predictor = loadModel(cls_model_file, std::stoi(num_threads));
std::vector<double> det_time_info = {0, 0, 0};
std::vector<double> rec_time_info = {0, 0, 0};
auto det_predictor = loadModel(det_model_file);
auto rec_predictor = loadModel(rec_model_file);
auto cls_predictor = loadModel(cls_model_file);
for (int i = 0; i < cv_all_img_names.size(); ++i) {
std::cout << "The predict img: " << cv_all_img_names[i] << std::endl;
cv::Mat srcimg = cv::imread(cv_all_img_names[i], cv::IMREAD_COLOR);
if (!srcimg.data) {
std::cerr << "[ERROR] image read failed! image path: " << cv_all_img_names[i] << std::endl;
exit(1);
}
std::vector<double> det_times;
auto boxes = RunDetModel(det_predictor, srcimg, Config, &det_times);
std::vector<std::string> rec_text;
std::vector<float> rec_text_score;
std::vector<double> rec_times;
RunRecModel(boxes, srcimg, rec_predictor, rec_text, rec_text_score,
charactor_dict, cls_predictor, use_direction_classify, &rec_times);
//// visualization
auto img_vis = Visualization(srcimg, boxes);
//// print recognized text
for (int i = 0; i < rec_text.size(); i++) {
std::cout << i << "\t" << rec_text[i] << "\t" << rec_text_score[i]
<< std::endl;
}
det_time_info[0] += det_times[0];
det_time_info[1] += det_times[1];
det_time_info[2] += det_times[2];
rec_time_info[0] += rec_times[0];
rec_time_info[1] += rec_times[1];
rec_time_info[2] += rec_times[2];
}
if (strcmp(argv[12], "True") == 0) {
AutoLogger autolog_det(det_model_file,
runtime_device,
std::stoi(num_threads),
std::stoi(batchsize),
"dynamic",
precision,
det_time_info,
cv_all_img_names.size());
AutoLogger autolog_rec(rec_model_file,
runtime_device,
std::stoi(num_threads),
std::stoi(batchsize),
"dynamic",
precision,
rec_time_info,
cv_all_img_names.size());
autolog_det.report();
std::cout << std::endl;
autolog_rec.report();
}
}
void det(int argc, char **argv) {
std::string det_model_file = argv[2];
std::string runtime_device = argv[3];
std::string precision = argv[4];
std::string num_threads = argv[5];
std::string batchsize = argv[6];
std::string img_dir = argv[7];
std::string det_config_path = argv[8];
if (strcmp(argv[4], "FP32") != 0 && strcmp(argv[4], "INT8") != 0) {
std::cerr << "Only support FP32 or INT8." << std::endl;
exit(1);
}
std::vector<cv::String> cv_all_img_names;
cv::glob(img_dir, cv_all_img_names);
//// load config from txt file
auto Config = LoadConfigTxt(det_config_path);
auto det_predictor = loadModel(det_model_file, std::stoi(num_threads));
std::vector<double> time_info = {0, 0, 0};
for (int i = 0; i < cv_all_img_names.size(); ++i) {
std::cout << "The predict img: " << cv_all_img_names[i] << std::endl;
cv::Mat srcimg = cv::imread(cv_all_img_names[i], cv::IMREAD_COLOR);
if (!srcimg.data) {
std::cerr << "[ERROR] image read failed! image path: " << cv_all_img_names[i] << std::endl;
exit(1);
}
std::vector<double> times;
auto boxes = RunDetModel(det_predictor, srcimg, Config, &times);
//// visualization
auto img_vis = Visualization(srcimg, boxes);
std::cout << boxes.size() << " bboxes have detected:" << std::endl;
for (int i=0; i<boxes.size(); i++){
std::cout << "The " << i << " box:" << std::endl;
for (int j=0; j<4; j++){
for (int k=0; k<2; k++){
std::cout << boxes[i][j][k] << "\t";
}
}
std::cout << std::endl;
}
time_info[0] += times[0];
time_info[1] += times[1];
time_info[2] += times[2];
}
if (strcmp(argv[9], "True") == 0) {
AutoLogger autolog(det_model_file,
runtime_device,
std::stoi(num_threads),
std::stoi(batchsize),
"dynamic",
precision,
time_info,
cv_all_img_names.size());
autolog.report();
}
}
void rec(int argc, char **argv) {
std::string rec_model_file = argv[2];
std::string runtime_device = argv[3];
std::string precision = argv[4];
std::string num_threads = argv[5];
std::string batchsize = argv[6];
std::string img_dir = argv[7];
std::string dict_path = argv[8];
if (strcmp(argv[4], "FP32") != 0 && strcmp(argv[4], "INT8") != 0) {
std::cerr << "Only support FP32 or INT8." << std::endl;
exit(1);
}
std::vector<cv::String> cv_all_img_names;
cv::glob(img_dir, cv_all_img_names);
auto charactor_dict = ReadDict(dict_path);
charactor_dict.insert(charactor_dict.begin(), "#"); // blank char for ctc
charactor_dict.push_back(" ");
cv::Mat srcimg = cv::imread(img_path, cv::IMREAD_COLOR);
auto boxes = RunDetModel(det_predictor, srcimg, Config);
auto rec_predictor = loadModel(rec_model_file, std::stoi(num_threads));
std::vector<std::string> rec_text;
std::vector<float> rec_text_score;
std::shared_ptr<PaddlePredictor> cls_predictor;
RunRecModel(boxes, srcimg, rec_predictor, rec_text, rec_text_score,
charactor_dict, cls_predictor, use_direction_classify);
std::vector<double> time_info = {0, 0, 0};
for (int i = 0; i < cv_all_img_names.size(); ++i) {
std::cout << "The predict img: " << cv_all_img_names[i] << std::endl;
cv::Mat srcimg = cv::imread(cv_all_img_names[i], cv::IMREAD_COLOR);
auto end = std::chrono::system_clock::now();
auto duration =
std::chrono::duration_cast<std::chrono::microseconds>(end - start);
if (!srcimg.data) {
std::cerr << "[ERROR] image read failed! image path: " << cv_all_img_names[i] << std::endl;
exit(1);
}
//// visualization
auto img_vis = Visualization(srcimg, boxes);
int width = srcimg.cols;
int height = srcimg.rows;
std::vector<int> upper_left = {0, 0};
std::vector<int> upper_right = {width, 0};
std::vector<int> lower_right = {width, height};
std::vector<int> lower_left = {0, height};
std::vector<std::vector<int>> box = {upper_left, upper_right, lower_right, lower_left};
std::vector<std::vector<std::vector<int>>> boxes = {box};
std::vector<std::string> rec_text;
std::vector<float> rec_text_score;
std::vector<double> times;
RunRecModel(boxes, srcimg, rec_predictor, rec_text, rec_text_score,
charactor_dict, cls_predictor, 0, &times);
//// print recognized text
for (int i = 0; i < rec_text.size(); i++) {
std::cout << i << "\t" << rec_text[i] << "\t" << rec_text_score[i]
<< std::endl;
}
time_info[0] += times[0];
time_info[1] += times[1];
time_info[2] += times[2];
}
// TODO: support autolog
if (strcmp(argv[9], "True") == 0) {
AutoLogger autolog(rec_model_file,
runtime_device,
std::stoi(num_threads),
std::stoi(batchsize),
"dynamic",
precision,
time_info,
cv_all_img_names.size());
autolog.report();
}
}
int main(int argc, char **argv) {
check_params(argc, argv);
std::cout << "mode: " << argv[1] << endl;
//// print recognized text
for (int i = 0; i < rec_text.size(); i++) {
std::cout << i << "\t" << rec_text[i] << "\t" << rec_text_score[i]
<< std::endl;
if (strcmp(argv[1], "system") == 0) {
system(argv);
}
std::cout << "花费了"
<< double(duration.count()) *
std::chrono::microseconds::period::num /
std::chrono::microseconds::period::den
<< "秒" << std::endl;
if (strcmp(argv[1], "det") == 0) {
det(argc, argv);
}
if (strcmp(argv[1], "rec") == 0) {
rec(argc, argv);
}
return 0;
}
\ No newline at end of file
}
# paddle2onnx 模型转化与预测
本章节介绍 PaddleOCR 模型如何转化为 ONNX 模型,并基于 ONNX 引擎预测。
## 1. 环境准备
需要准备 Paddle2ONNX 模型转化环境,和 ONNX 模型预测环境
### Paddle2ONNX
Paddle2ONNX 支持将 PaddlePaddle 模型格式转化到 ONNX 模型格式,算子目前稳定支持导出 ONNX Opset 9~11,部分Paddle算子支持更低的ONNX Opset转换。
更多细节可参考 [Paddle2ONNX](https://github.com/PaddlePaddle/Paddle2ONNX/blob/develop/README_zh.md)
- 安装 Paddle2ONNX
```
python3.7 -m pip install paddle2onnx
```
- 安装 ONNX
```
# 建议安装 1.4.0 版本,可根据环境更换版本号
python3.7 -m pip install onnxruntime==1.4.0
```
## 2. 模型转换
- Paddle 模型下载
有两种方式获取Paddle静态图模型:在 [model_list](../../doc/doc_ch/models_list.md) 中下载PaddleOCR提供的预测模型;
参考[模型导出说明](../../doc/doc_ch/inference.md#训练模型转inference模型)把训练好的权重转为 inference_model。
以 ppocr 检测模型为例:
```
wget -nc -P ./inference https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_det_infer.tar
cd ./inference && tar xf ch_ppocr_mobile_v2.0_det_infer.tar && cd ..
```
- 模型转换
使用 Paddle2ONNX 将Paddle静态图模型转换为ONNX模型格式:
```
paddle2onnx --model_dir=./inference/ch_ppocr_mobile_v2.0_det_infer/ \
--model_filename=inference.pdmodel \
--params_filename=inference.pdiparams \
--save_file=./inference/det_mobile_onnx/model.onnx \
--opset_version=10 \
--enable_onnx_checker=True
```
执行完毕后,ONNX 模型会被保存在 `./inference/det_mobile_onnx/` 路径下
* 注意:以下几个模型暂不支持转换为 ONNX 模型:
NRTR、SAR、RARE、SRN
## 3. onnx 预测
以检测模型为例,使用 ONNX 预测可执行如下命令:
```
python3.7 ../../tools/infer/predict_det.py --use_gpu=False --use_onnx=True \
--det_model_dir=./inference/det_mobile_onnx/model.onnx \
--image_dir=../../doc/imgs/1.jpg
```
执行命令后在终端会打印出预测的检测框坐标,并在 `./inference_results/` 下保存可视化结果。
```
root INFO: 1.jpg [[[291, 295], [334, 292], [348, 844], [305, 847]], [[344, 296], [379, 294], [387, 669], [353, 671]]]
The predict time of ../../doc/imgs/1.jpg: 0.06162881851196289
The visualized image saved in ./inference_results/det_res_1.jpg
```
* 注意:ONNX暂时不支持变长预测,需要将输入resize到固定输入,预测结果可能与直接使用Paddle预测有细微不同。
......@@ -114,7 +114,7 @@ The recognition model is the same.
git clone https://github.com/PaddlePaddle/PaddleOCR
# Enter the working directory
cd PaddleOCR/deploy/pdserver/
cd PaddleOCR/deploy/pdserving/
```
The pdserver directory contains the code to start the pipeline service and send prediction requests, including:
......
......@@ -112,7 +112,7 @@ python3 -m paddle_serving_client.convert --dirname ./ch_ppocr_mobile_v2.0_rec_in
git clone https://github.com/PaddlePaddle/PaddleOCR
# 进入到工作目录
cd PaddleOCR/deploy/pdserver/
cd PaddleOCR/deploy/pdserving/
```
pdserver目录包含启动pipeline服务和发送预测请求的代码,包括:
```
......@@ -206,7 +206,7 @@ pip3 install paddle-serving-app==0.3.1
1. 启动服务端程序
```
cd win
cd win
python3 ocr_web_server.py gpu(使用gpu方式)
或者
python3 ocr_web_server.py cpu(使用cpu方式)
......
# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# pylint: disable=doc-string-missing
from paddle_serving_client import Client
import sys
import numpy as np
import base64
import os
import cv2
from paddle_serving_app.reader import Sequential, URL2Image, ResizeByFactor
from paddle_serving_app.reader import Div, Normalize, Transpose
from ocr_reader import OCRReader
client = Client()
# TODO:load_client need to load more than one client model.
# this need to figure out some details.
client.load_client_config(sys.argv[1:])
client.connect(["127.0.0.1:9293"])
import paddle
test_img_dir = "test_img/"
ocr_reader = OCRReader(char_dict_path="../../ppocr/utils/ppocr_keys_v1.txt")
def cv2_to_base64(image):
return base64.b64encode(image).decode(
'utf8') #data.tostring()).decode('utf8')
for img_file in os.listdir(test_img_dir):
with open(os.path.join(test_img_dir, img_file), 'rb') as file:
image_data = file.read()
image = cv2_to_base64(image_data)
res_list = []
#print(image)
fetch_map = client.predict(
feed={"x": image}, fetch=["save_infer_model/scale_0.tmp_1"], batch=True)
print("fetrch map:", fetch_map)
one_batch_res = ocr_reader.postprocess(fetch_map, with_score=True)
for res in one_batch_res:
res_list.append(res[0])
res = {"res": str(res_list)}
print(res)
......@@ -18,13 +18,19 @@ import json
import base64
import os
import argparse
parser = argparse.ArgumentParser(description="args for paddleserving")
parser.add_argument("--image_dir", type=str, default="../../doc/imgs/")
args = parser.parse_args()
def cv2_to_base64(image):
return base64.b64encode(image).decode('utf8')
url = "http://127.0.0.1:9998/ocr/prediction"
test_img_dir = "../../doc/imgs/"
test_img_dir = args.image_dir
for idx, img_file in enumerate(os.listdir(test_img_dir)):
with open(os.path.join(test_img_dir, img_file), 'rb') as file:
image_data1 = file.read()
......@@ -36,5 +42,4 @@ for idx, img_file in enumerate(os.listdir(test_img_dir)):
r = requests.post(url=url, data=json.dumps(data))
print(r.json())
test_img_dir = "../../doc/imgs/"
print("==> total number of test imgs: ", len(os.listdir(test_img_dir)))
......@@ -30,12 +30,17 @@ def cv2_to_base64(image):
return base64.b64encode(image).decode('utf8')
test_img_dir = "imgs/"
import argparse
parser = argparse.ArgumentParser(description="args for paddleserving")
parser.add_argument("--image_dir", type=str, default="../../doc/imgs/")
args = parser.parse_args()
test_img_dir = args.image_dir
for img_file in os.listdir(test_img_dir):
with open(os.path.join(test_img_dir, img_file), 'rb') as file:
image_data = file.read()
image = cv2_to_base64(image_data)
for i in range(1):
ret = client.predict(feed_dict={"image": image}, fetch=["res"])
print(ret)
for i in range(1):
ret = client.predict(feed_dict={"image": image}, fetch=["res"])
print(ret)
......@@ -30,7 +30,7 @@ from ppocr.modeling.architectures import build_model
from ppocr.postprocess import build_post_process
from ppocr.metrics import build_metric
from ppocr.utils.save_load import init_model
from ppocr.utils.save_load import load_model
import tools.program as program
......@@ -89,7 +89,7 @@ def main(config, device, logger, vdl_writer):
logger.info(f"FLOPs after pruning: {flops}")
# load pretrain model
pre_best_model_dict = init_model(config, model, logger, None)
load_model(config, model)
metric = program.eval(model, valid_dataloader, post_process_class,
eval_class)
logger.info(f"metric['hmean']: {metric['hmean']}")
......
......@@ -32,7 +32,7 @@ from ppocr.losses import build_loss
from ppocr.optimizer import build_optimizer
from ppocr.postprocess import build_post_process
from ppocr.metrics import build_metric
from ppocr.utils.save_load import init_model
from ppocr.utils.save_load import load_model
import tools.program as program
dist.get_world_size()
......@@ -94,7 +94,7 @@ def main(config, device, logger, vdl_writer):
# build metric
eval_class = build_metric(config['Metric'])
# load pretrain model
pre_best_model_dict = init_model(config, model, logger, optimizer)
pre_best_model_dict = load_model(config, model, optimizer)
logger.info('train dataloader has {} iters, valid dataloader has {} iters'.
format(len(train_dataloader), len(valid_dataloader)))
......
......@@ -28,7 +28,7 @@ from paddle.jit import to_static
from ppocr.modeling.architectures import build_model
from ppocr.postprocess import build_post_process
from ppocr.utils.save_load import init_model
from ppocr.utils.save_load import load_model
from ppocr.utils.logging import get_logger
from tools.program import load_config, merge_config, ArgsParser
from ppocr.metrics import build_metric
......@@ -101,7 +101,7 @@ def main():
quanter = QAT(config=quant_config)
quanter.quantize(model)
init_model(config, model)
load_model(config, model)
model.eval()
# build metric
......
......@@ -37,7 +37,7 @@ from ppocr.losses import build_loss
from ppocr.optimizer import build_optimizer
from ppocr.postprocess import build_post_process
from ppocr.metrics import build_metric
from ppocr.utils.save_load import init_model
from ppocr.utils.save_load import load_model
import tools.program as program
from paddleslim.dygraph.quant import QAT
......@@ -137,7 +137,7 @@ def main(config, device, logger, vdl_writer):
# build metric
eval_class = build_metric(config['Metric'])
# load pretrain model
pre_best_model_dict = init_model(config, model, logger, optimizer)
pre_best_model_dict = load_model(config, model, optimizer)
logger.info('train dataloader has {} iters, valid dataloader has {} iters'.
format(len(train_dataloader), len(valid_dataloader)))
......
......@@ -37,7 +37,7 @@ from ppocr.losses import build_loss
from ppocr.optimizer import build_optimizer
from ppocr.postprocess import build_post_process
from ppocr.metrics import build_metric
from ppocr.utils.save_load import init_model
from ppocr.utils.save_load import load_model
import tools.program as program
import paddleslim
from paddleslim.dygraph.quant import QAT
......
<a name="算法介绍"></a>
## 算法介绍
# 两阶段算法
- [两阶段算法](#-----)
* [1. 算法介绍](#1)
+ [1.1 文本检测算法](#11)
+ [1.2 文本识别算法](#12)
* [2. 模型训练](#2)
* [3. 模型推理](#3)
<a name="1"></a>
## 1. 算法介绍
本文给出了PaddleOCR已支持的文本检测算法和文本识别算法列表,以及每个算法在**英文公开数据集**上的模型和指标,主要用于算法简介和算法性能对比,更多包括中文在内的其他数据集上的模型请参考[PP-OCR v2.0 系列模型下载](./models_list.md)
- [1.文本检测算法](#文本检测算法)
- [2.文本识别算法](#文本识别算法)
<a name="11"></a>
<a name="文本检测算法"></a>
### 1.文本检测算法
### 1.1 文本检测算法
PaddleOCR开源的文本检测算法列表:
- [x] DB([paper]( https://arxiv.org/abs/1911.08947))(ppocr推荐)
- [x] EAST([paper](https://arxiv.org/abs/1704.03155))
- [x] SAST([paper](https://arxiv.org/abs/1908.05498))
- [x] DB([paper]( https://arxiv.org/abs/1911.08947)) [2](ppocr推荐)
- [x] EAST([paper](https://arxiv.org/abs/1704.03155))[1]
- [x] SAST([paper](https://arxiv.org/abs/1908.05498))[4]
- [x] PSENet([paper](https://arxiv.org/abs/1903.12473v2)
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|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)|
|PSE|ResNet50_vd|85.81%|79.53%|82.55%|[下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.1/en_det/det_r50_vd_pse_v2.0_train.tar)|
|PSE|MobileNetV3|82.20%|70.48%|75.89%|[下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.1/en_det/det_mv3_pse_v2.0_train.tar)|
|EAST|ResNet50_vd|85.80%|86.71%|86.25%|[训练模型](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%|[训练模型](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%|[训练模型](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%|[训练模型](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%|[训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/det_r50_vd_sast_icdar15_v2.0_train.tar)|
|PSE|ResNet50_vd|85.81%|79.53%|82.55%|[训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.1/en_det/det_r50_vd_pse_v2.0_train.tar)|
|PSE|MobileNetV3|82.20%|70.48%|75.89%|[训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.1/en_det/det_mv3_pse_v2.0_train.tar)|
在Total-text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|
| --- | --- | --- | --- | --- | --- |
|SAST|ResNet50_vd|89.63%|78.44%|83.66%|[下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/det_r50_vd_sast_totaltext_v2.0_train.tar)|
|SAST|ResNet50_vd|89.63%|78.44%|83.66%|[训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/det_r50_vd_sast_totaltext_v2.0_train.tar)|
**说明:** SAST模型训练额外加入了icdar2013、icdar2017、COCO-Text、ArT等公开数据集进行调优。PaddleOCR用到的经过整理格式的英文公开数据集下载:
* [百度云地址](https://pan.baidu.com/s/12cPnZcVuV1zn5DOd4mqjVw) (提取码: 2bpi)
* [Google Drive下载地址](https://drive.google.com/drive/folders/1ll2-XEVyCQLpJjawLDiRlvo_i4BqHCJe?usp=sharing)
PaddleOCR文本检测算法的训练和使用请参考文档教程中[模型训练/评估中的文本检测部分](./detection.md)
<a name="12"></a>
<a name="文本识别算法"></a>
### 2.文本识别算法
### 1.2 文本识别算法
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))(ppocr推荐)
- [x] Rosetta([paper](https://arxiv.org/abs/1910.05085))
- [x] STAR-Net([paper](http://www.bmva.org/bmvc/2016/papers/paper043/index.html))
- [x] RARE([paper](https://arxiv.org/abs/1603.03915v1))
- [x] SRN([paper](https://arxiv.org/abs/2003.12294))
- [x] NRTR([paper](https://arxiv.org/abs/1806.00926v2))
- [x] CRNN([paper](https://arxiv.org/abs/1507.05717))[7](ppocr推荐)
- [x] Rosetta([paper](https://arxiv.org/abs/1910.05085))[10]
- [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))[13]
- [x] SAR([paper](https://arxiv.org/abs/1811.00751v2))
- [x] SEED([paper](https://arxiv.org/pdf/2005.10977.pdf))
- [x] SEED([paper](https://arxiv.org/pdf/2005.10977.pdf))
参考[DTRB](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:
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)|
|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) |
|SAR|Resnet31| 87.2% | rec_r31_sar | [下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.1/rec/rec_r31_sar_train.tar) |
|SEED| Aster_Resnet | 85.2% | rec_resnet_stn_bilstm_att | [下载链接](https://paddleocr.bj.bcebos.com/dygraph_v2.1/rec/rec_resnet_stn_bilstm_att.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)
|Rosetta|Resnet34_vd|80.9%|rec_r34_vd_none_none_ctc|[训练模型](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|[训练模型](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|[训练模型](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|[训练模型](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|[训练模型](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|[训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_mv3_tps_bilstm_ctc_v2.0_train.tar)|
|RARE|MobileNetV3|82.5%|rec_mv3_tps_bilstm_att |[训练模型](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 |[训练模型](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 | [训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_r50_vd_srn_train.tar) |
|NRTR|NRTR_MTB| 84.3% | rec_mtb_nrtr | [训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.0/en/rec_mtb_nrtr_train.tar) |
|SAR|Resnet31| 87.2% | rec_r31_sar | [训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.1/rec/rec_r31_sar_train.tar) |
|SEED|Aster_Resnet| 85.2% | rec_resnet_stn_bilstm_att | [训练模型](https://paddleocr.bj.bcebos.com/dygraph_v2.1/rec/rec_resnet_stn_bilstm_att.tar) |
<a name="2"></a>
## 2. 模型训练
PaddleOCR文本检测算法的训练和使用请参考文档教程中[模型训练/评估中的文本检测部分](./detection.md)。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)
<a name="3"></a>
## 3. 模型推理
上述模型中除PP-OCR系列模型以外,其余模型仅支持基于Python引擎的推理,具体内容可参考[基于Python预测引擎推理](./inference.md)
......@@ -11,7 +11,7 @@
## 1. 方法介绍
文本方向分类器主要用于图片非0度的场景下,在这种场景下需要对图片里检测到的文本行进行一个转正的操作。在PaddleOCR系统内,
文字检测之后得到的文本行图片经过仿射变换之后送入识别模型,此时只需要对文字进行一个0和180度的角度分类,因此PaddleOCR内置的
字角度分类器**只支持了0和180度的分类**。如果想支持更多角度,可以自己修改算法进行支持。
本方向分类器**只支持了0和180度的分类**。如果想支持更多角度,可以自己修改算法进行支持。
0和180度数据样本例子:
......@@ -72,8 +72,6 @@ train/cls/train/word_002.jpg 180
<a name="启动训练"></a>
## 3. 启动训练
### 启动训练
将准备好的txt文件和图片文件夹路径分别写入配置文件的 `Train/Eval.dataset.label_file_list``Train/Eval.dataset.data_dir` 字段下,`Train/Eval.dataset.data_dir`字段下的路径和文件里记载的图片名构成了图片的绝对路径。
PaddleOCR提供了训练脚本、评估脚本和预测脚本。
......
......@@ -36,11 +36,10 @@
| pretrained_model | 设置加载预训练模型路径 | ./pretrain_models/CRNN/best_accuracy | \ |
| checkpoints | 加载模型参数路径 | None | 用于中断后加载参数继续训练 |
| use_visualdl | 设置是否启用visualdl进行可视化log展示 | False | [教程地址](https://www.paddlepaddle.org.cn/paddle/visualdl) |
| infer_img | 设置预测图像路径或文件夹路径 | ./infer_img | \|
| character_dict_path | 设置字典路径 | ./ppocr/utils/ppocr_keys_v1.txt | \ |
| infer_img | 设置预测图像路径或文件夹路径 | ./infer_img | \||
| character_dict_path | 设置字典路径 | ./ppocr/utils/ppocr_keys_v1.txt | 如果为空,则默认使用小写字母+数字作为字典 |
| max_text_length | 设置文本最大长度 | 25 | \ |
| character_type | 设置字符类型 | ch | en/ch, en时将使用默认dict,ch时使用自定义dict|
| use_space_char | 设置是否识别空格 | True | 仅在 character_type=ch 时支持空格 |
| use_space_char | 设置是否识别空格 | True | \| |
| label_list | 设置方向分类器支持的角度 | ['0','180'] | 仅在方向分类器中生效 |
| save_res_path | 设置检测模型的结果保存地址 | ./output/det_db/predicts_db.txt | 仅在检测模型中生效 |
......@@ -191,7 +190,6 @@ PaddleOCR目前已支持80种(除中文外)语种识别,`configs/rec/multi
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:
......@@ -212,17 +210,17 @@ PaddleOCR目前已支持80种(除中文外)语种识别,`configs/rec/multi
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 |
| 配置文件 | 算法名称 | backbone | trans | seq | pred | language |
| :--------: | :-------: | :-------: | :-------: | :-----: | :-----: | :-----: |
| rec_chinese_cht_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 中文繁体 |
| rec_en_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 英语(区分大小写) |
| rec_french_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 法语 |
| rec_ger_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 德语 |
| rec_japan_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 日语 |
| rec_korean_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 韩语 |
| rec_latin_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 拉丁字母 |
| rec_arabic_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 阿拉伯字母 |
| rec_cyrillic_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 斯拉夫字母 |
| rec_devanagari_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 梵文字母 |
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)
# 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)
<a name="1-----"></a>
# 1. 文字检测
# 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。
- [1. 准备数据和模型](#1--------)
* [1.1 数据准备](#11-----)
* [1.2 下载预训练模型](#12--------)
- [2. 开始训练](#2-----)
* [2.1 启动训练](#21-----)
* [2.2 断点训练](#22-----)
* [2.3 更换Backbone 训练](#23---backbone---)
- [3. 模型评估与预测](#3--------)
* [3.1 指标评估](#31-----)
* [3.2 测试检测效果](#32-------)
- [4. 模型导出与预测](#4--------)
- [5. FAQ](#5-faq)
<a name="1--------"></a>
# 1. 准备数据和模型
<a name="11-----"></a>
## 1.1 数据准备
......@@ -83,8 +85,11 @@ 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/ResNet50_vd_ssld_pretrained.pdparams
```
<a name="13-----"></a>
## 1.3 启动训练
<a name="2-----"></a>
# 2. 开始训练
<a name="21-----"></a>
## 2.1 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*
......@@ -96,6 +101,10 @@ python3 tools/train.py -c configs/det/det_mv3_db.yml \
# 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.pretrained_model=./pretrain_models/MobileNetV3_large_x0_5_pretrained
# 多机多卡训练,通过 --ips 参数设置使用的机器IP地址,通过 --gpus 参数设置使用的GPU ID
python3 -m paddle.distributed.launch --ips="xx.xx.xx.xx,xx.xx.xx.xx" --gpus '0,1,2,3' tools/train.py -c configs/det/det_mv3_db.yml \
-o Global.pretrained_model=./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。
......@@ -106,8 +115,17 @@ 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
```
<a name="14-----"></a>
## 1.4 断点训练
**注意:** 采用多机多卡训练时,需要替换上面命令中的ips值为您机器的地址,机器之间需要能够相互ping通。另外,训练时需要在多个机器上分别启动命令。查看机器ip地址的命令为`ifconfig`
如果您想进一步加快训练速度,可以使用[自动混合精度训练](https://www.paddlepaddle.org.cn/documentation/docs/zh/guides/01_paddle2.0_introduction/basic_concept/amp_cn.html), 以单机单卡为例,命令如下:
```shell
python3 tools/train.py -c configs/det/det_mv3_db.yml \
-o Global.pretrained_model=./pretrain_models/MobileNetV3_large_x0_5_pretrained \
Global.use_amp=True Global.scale_loss=1024.0 Global.use_dynamic_loss_scaling=True
```
<a name="22-----"></a>
## 2.2 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
......@@ -116,8 +134,8 @@ python3 tools/train.py -c configs/det/det_mv3_db.yml -o Global.checkpoints=./you
**注意**:`Global.checkpoints`的优先级高于`Global.pretrained_model`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.pretrained_model`指定的模型。
<a name="15---backbone---"></a>
## 1.5 更换Backbone 训练
<a name="23---backbone---"></a>
## 2.3 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。
......@@ -164,8 +182,11 @@ args1: args1
**注意**:如果要更换网络的其他模块,可以参考[文档](./add_new_algorithm.md)。
<a name="16-----"></a>
## 1.6 指标评估
<a name="3--------"></a>
# 3. 模型评估与预测
<a name="31-----"></a>
## 3.1 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)。
......@@ -177,8 +198,8 @@ python3 tools/eval.py -c configs/det/det_mv3_db.yml -o Global.checkpoints="{pat
* 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
<a name="17-------"></a>
## 1.7 测试检测效果
<a name="32-------"></a>
## 3.2 测试检测效果
测试单张图像的检测效果
```shell
......@@ -195,8 +216,8 @@ python3 tools/infer_det.py -c configs/det/det_mv3_db.yml -o Global.infer_img="./
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
<a name="4--------"></a>
# 4. 模型导出与预测
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。
......@@ -218,10 +239,11 @@ python3 tools/infer/predict_det.py --det_algorithm="DB" --det_model_dir="./outpu
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
<a name="5-faq"></a>
# 5. 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。
# Enhanced CTC Loss
在OCR识别中, CRNN是一种在工业界广泛使用的文字识别算法。 在训练阶段,其采用CTCLoss来计算网络损失; 在推理阶段,其采用CTCDecode来获得解码结果。虽然CRNN算法在实际业务中被证明能够获得很好的识别效果, 然而用户对识别准确率的要求却是无止境的,如何进一步提升文字识别的准确率呢? 本文以CTCLoss为切人点,分别从难例挖掘、 多任务学习、 Metric Learning 3个不同的角度探索了CTCLoss的改进融合方案,提出了EnhancedCTCLoss,其包括如下3个组成部分: Focal-CTC Loss,A-CTC Loss, C-CTC Loss。
## 1. Focal-CTC Loss
Focal Loss 出自论文《Focal Loss for Dense Object Detection》, 该loss最先提出的时候主要是为了解决one-stage目标检测中正负样本比例严重失衡的问题。该损失函数降低了大量简单负样本在训练中所占的权重,也可理解为一种困难样本挖掘。
其损失函数形式如下:
<div align="center">
<img src="./focal_loss_formula.png" width = "600" />
</div>
其中, y' 是经过激活函数的输出,取值在0-1之间。其在原始的交叉熵损失的基础上加了一个调制系数(1 – y’)^ &gamma;和平衡因子&alpha;。 当&alpha; = 1,y=1时,其损失函数与交叉熵损失的对比如下图所示:
<div align="center">
<img src="./focal_loss_image.png" width = "600" />
</div>
从上图可以看到, 当&gamma;> 0时,调整系数(1-y’)^&gamma; 赋予易分类样本损失一个更小的权重,使得网络更关注于困难的、错分的样本。 调整因子&gamma;用于调节简单样本权重降低的速率,当&gamma;为0时即为交叉熵损失函数,当&gamma;增加时,调整因子的影响也会随之增大。实验发现&gamma;为2是最优。平衡因子&alpha;用来平衡正负样本本身的比例不均,文中&alpha;取0.25。
对于经典的CTC算法,假设某个特征序列(f<sub>1</sub>, f<sub>2</sub>, ......f<sub>t</sub>), 经过CTC解码之后结果等于label的概率为y’, 则CTC解码结果不为label的概率即为(1-y’);不难发现, CTCLoss值和y’有如下关系:
<div align="center">
<img src="./equation_ctcloss.png" width = "250" />
</div>
结合Focal Loss的思想,赋予困难样本较大的权重,简单样本较小的权重,可以使网络更加聚焦于对困难样本的挖掘,进一步提升识别的准确率,由此我们提出了Focal-CTC Loss; 其定义如下所示:
<div align="center">
<img src="./equation_focal_ctc.png" width = "500" />
</div>
实验中,&gamma;取值为2, &alpha;= 1, 具体实现见: [rec_ctc_loss.py](../../ppocr/losses/rec_ctc_loss.py)
## 2. A-CTC Loss
A-CTC Loss是CTC Loss + ACE Loss的简称。 其中ACE Loss出自论文< Aggregation Cross-Entropy for Sequence Recognition>. ACE Loss相比于CTCLoss,主要有如下两点优势:
+ ACE Loss能够解决2-D文本的识别问题; CTCLoss只能够处理1-D文本
+ ACE Loss 在时间复杂度和空间复杂度上优于CTC loss
前人总结的OCR识别算法的优劣如下图所示:
<div align="center">
<img src="./rec_algo_compare.png" width = "1000" />
</div>
虽然ACELoss确实如上图所说,可以处理2D预测,在内存占用及推理速度方面具备优势,但在实践过程中,我们发现单独使用ACE Loss, 识别效果并不如CTCLoss. 因此,我们尝试将CTCLoss和ACELoss进行结合,同时以CTCLoss为主,将ACELoss 定位为一个辅助监督loss。 这一尝试收到了效果,在我们内部的实验数据集上,相比单独使用CTCLoss,识别准确率可以提升1%左右。
A_CTC Loss定义如下:
<div align="center">
<img src="./equation_a_ctc.png" width = "300" />
</div>
实验中,λ = 0.1. ACE loss实现代码见: [ace_loss.py](../../ppocr/losses/ace_loss.py)
## 3. C-CTC Loss
C-CTC Loss是CTC Loss + Center Loss的简称。 其中Center Loss出自论文 < A Discriminative Feature Learning Approach for Deep Face Recognition>. 最早用于人脸识别任务,用于增大类间距离,减小类内距离, 是Metric Learning领域一种较早的、也比较常用的一种算法。
在中文OCR识别任务中,通过对badcase分析, 我们发现中文识别的一大难点是相似字符多,容易误识。 由此我们想到是否可以借鉴Metric Learing的想法, 增大相似字符的类间距,从而提高识别准确率。然而,MetricLearning主要用于图像识别领域,训练数据的标签为一个固定的值;而对于OCR识别来说,其本质上是一个序列识别任务,特征和label之间并不具有显式的对齐关系,因此两者如何结合依然是一个值得探索的方向。
通过尝试Arcmargin, Cosmargin等方法, 我们最终发现Centerloss 有助于进一步提升识别的准确率。C_CTC Loss定义如下:
<div align="center">
<img src="./equation_c_ctc.png" width = "300" />
</div>
实验中,我们设置λ=0.25. center_loss实现代码见: [center_loss.py](../../ppocr/losses/center_loss.py)
值得一提的是, 在C-CTC Loss中,选择随机初始化Center并不能够带来明显的提升. 我们的Center初始化方法如下:
+ 基于原始的CTCLoss, 训练得到一个网络N
+ 挑选出训练集中,识别完全正确的部分, 组成集合G
+ 将G中的每个样本送入网络,进行前向计算, 提取最后一个FC层的输入(即feature)及其经过argmax计算的结果(即index)之间的对应关系
+ 将相同index的feature进行聚合,计算平均值,得到各自字符的初始center.
以配置文件`configs/rec/ch_PP-OCRv2/ch_PP-OCRv2_rec.yml`为例, center提取命令如下所示:
```
python tools/export_center.py -c configs/rec/ch_PP-OCRv2/ch_PP-OCRv2_rec.yml -o Global.pretrained_model="./output/rec_mobile_pp-OCRv2/best_accuracy"
```
运行完后,会在PaddleOCR主目录下生成`train_center.pkl`.
## 4. 实验
对于上述的三种方案,我们基于百度内部数据集进行了训练、评测,实验情况如下表所示:
|algorithm| Focal_CTC | A_CTC | C-CTC |
|:------| :------| ------: | :------: |
|gain| +0.3% | +0.7% | +1.7% |
基于上述实验结论,我们在PP-OCRv2中,采用了C-CTC的策略。 值得一提的是,由于PP-OCRv2 处理的是6625个中文字符的识别任务,字符集比较大,形似字较多,所以在该任务上C-CTC 方案带来的提升较大。 但如果换做其他OCR识别任务,结论可能会有所不同。大家可以尝试Focal-CTC,A-CTC, C-CTC以及组合方案EnhancedCTC,相信会带来不同程度的提升效果。
统一的融合方案见如下文件: [rec_enhanced_ctc_loss.py](../../ppocr/losses/rec_enhanced_ctc_loss.py)
# 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。
推荐环境:
- PaddlePaddle >= 2.0.0 (2.1.2)
- python3.7
- CUDA10.1 / CUDA10.2
- CUDNN 7.6
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)
......@@ -123,13 +130,13 @@ Windows和Mac用户推荐使用Anaconda搭建Python环境,Linux用户建议使
# !! 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"
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
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 <<<
......@@ -294,11 +301,12 @@ 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 registry.baidubce.com/paddlepaddle/paddle:2.1.3-gpu-cuda10.2-cudnn7 /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
#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
# 如果是CUDA11+CUDNN8,推荐使用镜像registry.baidubce.com/paddlepaddle/paddle:2.1.3-gpu-cuda11.2-cudnn8
sudo nvidia-docker run --name ppocr -v $PWD:/paddle --shm-size=64G --network=host -it registry.baidubce.com/paddlepaddle/paddle:2.1.3-gpu-cuda10.2-cudnn7 /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
......@@ -321,8 +329,3 @@ 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。
......@@ -273,7 +273,7 @@ python3 tools/export_model.py -c configs/rec/rec_r34_vd_none_bilstm_ctc.yml -o G
CRNN 文本识别模型推理,可以执行如下命令:
```
python3 tools/infer/predict_rec.py --image_dir="./doc/imgs_words_en/word_336.png" --rec_model_dir="./inference/rec_crnn/" --rec_image_shape="3, 32, 100" --rec_char_type="en"
python3 tools/infer/predict_rec.py --image_dir="./doc/imgs_words_en/word_336.png" --rec_model_dir="./inference/rec_crnn/" --rec_image_shape="3, 32, 100" --rec_char_dict_path="./ppocr/utils/ic15_dict.txt"
```
![](../imgs_words_en/word_336.png)
......@@ -288,7 +288,7 @@ Predicts of ./doc/imgs_words_en/word_336.png:('super', 0.9999073)
- 训练时采用的图像分辨率不同,训练上述模型采用的图像分辨率是[3,32,100],而中文模型训练时,为了保证长文本的识别效果,训练时采用的图像分辨率是[3, 32, 320]。预测推理程序默认的的形状参数是训练中文采用的图像分辨率,即[3, 32, 320]。因此,这里推理上述英文模型时,需要通过参数rec_image_shape设置识别图像的形状。
- 字符列表,DTRB论文中实验只是针对26个小写英文本母和10个数字进行实验,总共36个字符。所有大小字符都转成了小写字符,不在上面列表的字符都忽略,认为是空格。因此这里没有输入字符字典,而是通过如下命令生成字典.因此在推理时需要设置参数rec_char_type,指定为英文"en"。
- 字符列表,DTRB论文中实验只是针对26个小写英文本母和10个数字进行实验,总共36个字符。所有大小字符都转成了小写字符,不在上面列表的字符都忽略,认为是空格。因此这里没有输入字符字典,而是通过如下命令生成字典.因此在推理时需要设置参数rec_char_dict_path,指定为英文字典"./ppocr/utils/ic15_dict.txt"。
```
self.character_str = "0123456789abcdefghijklmnopqrstuvwxyz"
......@@ -303,15 +303,15 @@ dict_character = list(self.character_str)
python3 tools/infer/predict_rec.py --image_dir="./doc/imgs_words_en/word_336.png" \
--rec_model_dir="./inference/srn/" \
--rec_image_shape="1, 64, 256" \
--rec_char_type="en" \
--rec_char_dict_path="./ppocr/utils/ic15_dict.txt" \
--rec_algorithm="SRN"
```
### 4. 自定义文本识别字典的推理
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`
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
```
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"
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_dict_path="your text dict path"
```
<a name="多语言模型的推理"></a>
......@@ -320,7 +320,7 @@ python3 tools/infer/predict_rec.py --image_dir="./doc/imgs_words_en/word_336.png
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"
python3 tools/infer/predict_rec.py --image_dir="./doc/imgs_words/korean/1.jpg" --rec_model_dir="./your inference model" --rec_char_dict_path="ppocr/utils/dict/korean_dict.txt" --vis_font_path="doc/fonts/korean.ttf"
```
![](../imgs_words/korean/1.jpg)
......@@ -388,7 +388,7 @@ python3 tools/infer/predict_system.py --image_dir="./doc/imgs/00018069.jpg" --de
下面给出基于EAST文本检测和STAR-Net文本识别执行命令:
```
python3 tools/infer/predict_system.py --image_dir="./doc/imgs_en/img_10.jpg" --det_model_dir="./inference/det_east/" --det_algorithm="EAST" --rec_model_dir="./inference/starnet/" --rec_image_shape="3, 32, 100" --rec_char_type="en"
python3 tools/infer/predict_system.py --image_dir="./doc/imgs_en/img_10.jpg" --det_model_dir="./inference/det_east/" --det_algorithm="EAST" --rec_model_dir="./inference/starnet/" --rec_image_shape="3, 32, 100" --rec_char_dict_path="./ppocr/utils/ic15_dict.txt"
```
执行命令后,识别结果图像如下:
......
# PP-OCR模型库快速推理
# 基于Python引擎的PP-OCR模型库推理
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>
......@@ -21,12 +18,15 @@
```
# 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/"
wget https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_det_infer.tar
tar xf ch_PP-OCRv2_det_infer.tar
python3 tools/infer/predict_det.py --image_dir="./doc/imgs/00018069.jpg" --det_model_dir="./ch_PP-OCRv2_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)
![](../imgs_results/det_res_00018069.jpg)
通过参数`limit_type``det_limit_side_len`来对图片的尺寸进行限制,
`limit_type`可选参数为[`max`, `min`],
......@@ -39,13 +39,13 @@ wget https://paddleocr.bj.bcebos.com/dygraph_v2.0/ch/ch_ppocr_mobile_v2.0_det_i
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
python3 tools/infer/predict_det.py --image_dir="./doc/imgs/1.jpg" --det_model_dir="./inference/ch_PP-OCRv2_det_infer/" --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
python3 tools/infer/predict_det.py --image_dir="./doc/imgs/1.jpg" --det_model_dir="./inference/ch_PP-OCRv2_det_infer/" --use_gpu=False
```
......@@ -62,12 +62,12 @@ python3 tools/infer/predict_det.py --image_dir="./doc/imgs/1.jpg" --det_model_di
```
# 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"
wget https://paddleocr.bj.bcebos.com/PP-OCRv2/chinese/ch_PP-OCRv2_rec_infer.tar
tar xf ch_PP-OCRv2_rec_infer.tar
python3 tools/infer/predict_rec.py --image_dir="./doc/imgs_words/ch/word_4.jpg" --rec_model_dir="./ch_PP-OCRv2_rec_infer/"
```
![](/Users/zhulingfeng01/OCR/PaddleOCR/doc/imgs_words/ch/word_4.jpg)
![](../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:
......@@ -79,14 +79,13 @@ Predicts of ./doc/imgs_words/ch/word_4.jpg:('实力活力', 0.98458153)
### 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:
如果您需要预测的是其他语言模型,可以在[此链接](./models_list.md#%E5%A4%9A%E8%AF%AD%E8%A8%80%E8%AF%86%E5%88%AB%E6%A8%A1%E5%9E%8B)中找到对应语言的inference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:
```
wget https://paddleocr.bj.bcebos.com/dygraph_v2.0/multilingual/korean_mobile_v2.0_rec_infer.tar
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)
![](../imgs_words/korean/1.jpg)
执行命令后,上图的预测结果为:
......@@ -107,7 +106,7 @@ 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)
![](../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:
......@@ -123,14 +122,13 @@ Predicts of ./doc/imgs_words/ch/word_4.jpg:['0', 0.9999982]
```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
python3 tools/infer/predict_system.py --image_dir="./doc/imgs/00018069.jpg" --det_model_dir="./inference/ch_PP-OCRv2_det_infer/" --cls_model_dir="./inference/cls/" --rec_model_dir="./inference/ch_PP-OCRv2_rec_infer/" --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
python3 tools/infer/predict_system.py --image_dir="./doc/imgs/00018069.jpg" --det_model_dir="./inference/ch_PP-OCRv2_det_infer/" --rec_model_dir="./inference/ch_PP-OCRv2_rec_infer/" --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
python3 tools/infer/predict_system.py --image_dir="./doc/imgs/00018069.jpg" --det_model_dir="./inference/ch_PP-OCRv2_det_infer/" --rec_model_dir="./inference/ch_PP-OCRv2_rec_infer/" --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)
![](../imgs_results/system_res_00018069.jpg)
# PP-OCR模型库
PP-OCR模型一节主要补充一些OCR模型的基本概念以及如何快速运用PP-OCR模型库中的模型。
本节包含两个部分,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[基于Python引擎的PP-OCR模型库推理](./inference_ppocr.md)中介绍PP-OCR模型库的使用方法,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。
## OCR模型列表(V2.1,2021年9月6日更新)
# OCR模型列表(V2.1,2021年9月6日更新)
> **说明**
> 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
......@@ -6,13 +6,13 @@
> 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)
- [二、文本识别模型](#文本识别模型)
- [1. 中文识别模型](#中文识别模型)
- [2. 英文识别模型](#英文识别模型)
- [3. 多语言识别模型](#多语言识别模型)
- [三、文本方向分类模型](#文本方向分类模型)
- [四、Paddle-Lite 模型](#Paddle-Lite模型)
- [1. 文本检测模型](#文本检测模型)
- [2. 文本识别模型](#文本识别模型)
- [2.1 中文识别模型](#中文识别模型)
- [2.2 英文识别模型](#英文识别模型)
- [2.3 多语言识别模型](#多语言识别模型)
- [3. 文本方向分类模型](#文本方向分类模型)
- [4. Paddle-Lite 模型](#Paddle-Lite模型)
PaddleOCR提供的可下载模型包括`推理模型``训练模型``预训练模型``slim模型`,模型区别说明如下:
......@@ -29,27 +29,28 @@ PaddleOCR提供的可下载模型包括`推理模型`、`训练模型`、`预训
<a name="文本检测模型"></a>
### 一、文本检测模型
## 1. 文本检测模型
|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_PP-OCRv2_det_slim|【最新】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-OCRv2_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|【最新】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-OCRv2_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)|
<a name="文本识别模型"></a>
### 二、文本识别模型
## 2. 文本识别模型
<a name="中文识别模型"></a>
#### 1. 中文识别模型
### 2.1 中文识别模型
|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_PP-OCRv2_rec_slim|【最新】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|【最新】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) |
......@@ -57,7 +58,7 @@ PaddleOCR提供的可下载模型包括`推理模型`、`训练模型`、`预训
**说明:** `训练模型`是基于预训练模型在真实数据与竖排合成文本数据上finetune得到的模型,在真实应用场景中有着更好的表现,`预训练模型`则是直接基于全量真实数据与合成数据训练得到,更适合用于在自己的数据集上finetune。
<a name="英文识别模型"></a>
#### 2. 英文识别模型
### 2.2 英文识别模型
|模型名称|模型简介|配置文件|推理模型大小|下载地址|
| --- | --- | --- | --- | --- |
......@@ -65,7 +66,7 @@ PaddleOCR提供的可下载模型包括`推理模型`、`训练模型`、`预训
|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)|2.6M|[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>
#### 3. 多语言识别模型(更多语言持续更新中...)
### 2.3 多语言识别模型(更多语言持续更新中...)
|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|
| --- | --- | --- | --- |--- | --- |
......@@ -86,7 +87,7 @@ PaddleOCR提供的可下载模型包括`推理模型`、`训练模型`、`预训
<a name="文本方向分类模型"></a>
### 三、文本方向分类模型
## 3. 文本方向分类模型
|模型名称|模型简介|配置文件|推理模型大小|下载地址|
| --- | --- | --- | --- | --- |
......@@ -95,7 +96,7 @@ PaddleOCR提供的可下载模型包括`推理模型`、`训练模型`、`预训
<a name="Paddle-Lite模型"></a>
### 四、Paddle-Lite 模型
## 4. 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|
|---|---|---|---|---|---|---|
......
......@@ -28,9 +28,9 @@ PGNet算法细节详见[论文](https://www.aaai.org/AAAI21Papers/AAAI-2885.Wang
### 性能指标
测试集: Total Text
#### 测试集: Total Text
测试环境: NVIDIA Tesla V100-SXM2-16GB
#### 测试环境: NVIDIA Tesla V100-SXM2-16GB
|PGNetA|det_precision|det_recall|det_f_score|e2e_precision|e2e_recall|e2e_f_score|FPS|下载|
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
......@@ -43,7 +43,7 @@ PGNet算法细节详见[论文](https://www.aaai.org/AAAI21Papers/AAAI-2885.Wang
<a name="环境配置"></a>
## 二、环境配置
请先参考[快速安装](./installation.md)配置PaddleOCR运行环境。
请先参考[《运行环境准备》](./environment.md)配置PaddleOCR运行环境,参考[《PaddleOCR全景图与项目克隆》](./paddleOCR_overview.md)克隆项目
<a name="快速使用"></a>
## 三、快速使用
......@@ -92,7 +92,7 @@ python3 tools/infer/predict_e2e.py --e2e_algorithm="PGNet" --image_dir="./doc/im
|- train.txt # total_text数据集的训练标注
```
total_text.txt标注文件格式如下,文件名和标注信息中间用"\t"分隔:
train.txt标注文件格式如下,文件名和标注信息中间用"\t"分隔:
```
" 图像文件名 json.dumps编码的图像标注信息"
rgb/img11.jpg [{"transcription": "ASRAMA", "points": [[214.0, 325.0], [235.0, 308.0], [259.0, 296.0], [286.0, 291.0], [313.0, 295.0], [338.0, 305.0], [362.0, 320.0], [349.0, 347.0], [330.0, 337.0], [310.0, 329.0], [290.0, 324.0], [269.0, 328.0], [249.0, 336.0], [231.0, 346.0]]}, {...}]
......
......@@ -47,10 +47,10 @@ cd /path/to/ppocr_img
<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。
* 检测+方向分类器+识别全流程:`--use_angle_cls true`设置使用方向分类器识别180度旋转文字,`--use_gpu false`设置不使用GPU
```bash
paddleocr --image_dir ./imgs/11.jpg --use_angle_cls true
paddleocr --image_dir ./imgs/11.jpg --use_angle_cls true --use_gpu false
```
结果是一个list,每个item包含了文本框,文字和识别置信度
......
......@@ -159,7 +159,6 @@ PaddleOCR内置了一部分字典,可以按需使用。
- 自定义字典
如需自定义dic文件,请在 `configs/rec/rec_icdar15_train.yml` 中添加 `character_dict_path` 字段, 指向您的字典路径。
并将 `character_type` 设置为 `ch`
<a name="支持空格"></a>
### 1.4 添加空格类别
......@@ -246,8 +245,6 @@ Global:
...
# 添加自定义字典,如修改字典请将路径指向新字典
character_dict_path: ppocr/utils/ppocr_keys_v1.txt
# 修改字符类型
character_type: ch
...
# 识别空格
use_space_char: True
......@@ -311,18 +308,18 @@ PaddleOCR目前已支持80种(除中文外)语种识别,`configs/rec/multi
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 |
| :--------: | :-------: | :-------: | :-------: | :-----: | :-----: | :-----: | :-----: |
| 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 |
| 配置文件 | 算法名称 | backbone | trans | seq | pred | language |
| :--------: | :-------: | :-------: | :-------: | :-----: | :-----: | :-----: |
| rec_chinese_cht_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 中文繁体 |
| rec_en_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 英语(区分大小写) |
| rec_french_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 法语 |
| rec_ger_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 德语 |
| rec_japan_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 日语 |
| rec_korean_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 韩语 |
| rec_latin_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 拉丁字母 |
| rec_arabic_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 阿拉伯字母 |
| rec_cyrillic_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 斯拉夫字母 |
| rec_devanagari_lite_train.yml | CRNN | Mobilenet_v3 small 0.5 | None | BiLSTM | ctc | 梵文字母 |
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)
......
......@@ -112,4 +112,14 @@
year={2016}
}
13.NRTR
@misc{sheng2019nrtr,
title={NRTR: A No-Recurrence Sequence-to-Sequence Model For Scene Text Recognition},
author={Fenfen Sheng and Zhineng Chen and Bo Xu},
year={2019},
eprint={1806.00926},
archivePrefix={arXiv},
primaryClass={cs.CV}
}
```
# 社区贡献说明
感谢大家长久以来对PaddleOCR的支持和关注,与广大开发者共同构建一个专业、和谐、相互帮助的开源社区是PaddleOCR的目标。本文档展示了已有的社区贡献、对于各类贡献说明、新的机会与流程,希望贡献流程更加高效、路径更加清晰。
PaddleOCR希望可以通过AI的力量助力任何一位有梦想的开发者实现自己的想法,享受创造价值带来的愉悦。
<a href="https://github.com/PaddlePaddle/PaddleOCR/graphs/contributors">
<img src="https://contrib.rocks/image?repo=PaddlePaddle/PaddleOCR" />
</a>
> 上图为PaddleOCR目前的Contributor,定期更新
## 1. 社区贡献
### 1.1 为PaddleOCR新增功能
- 非常感谢 [authorfu](https://github.com/authorfu) 贡献Android([#340](https://github.com/PaddlePaddle/PaddleOCR/pull/340))和[xiadeye](https://github.com/xiadeye) 贡献IOS的demo代码([#325](https://github.com/PaddlePaddle/PaddleOCR/pull/325))
- 非常感谢 [tangmq](https://gitee.com/tangmq) 给PaddleOCR增加Docker化部署服务,支持快速发布可调用的Restful API服务([#507](https://github.com/PaddlePaddle/PaddleOCR/pull/507))。
- 非常感谢 [lijinhan](https://github.com/lijinhan) 给PaddleOCR增加java SpringBoot 调用OCR Hubserving接口完成对OCR服务化部署的使用([#1027](https://github.com/PaddlePaddle/PaddleOCR/pull/1027))。
- 非常感谢 [Evezerest](https://github.com/Evezerest)[ninetailskim](https://github.com/ninetailskim)[edencfc](https://github.com/edencfc)[BeyondYourself](https://github.com/BeyondYourself)[1084667371](https://github.com/1084667371) 贡献了[PPOCRLabel](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.3/PPOCRLabel/README_ch.md) 的完整代码。
### 1.2 基于PaddleOCR的社区贡献
- 【最新】完整的C#版本标注工具 [FastOCRLabel](https://gitee.com/BaoJianQiang/FastOCRLabel) (@ [包建强](https://gitee.com/BaoJianQiang) )
- 通用型桌面级即时翻译工具 [DangoOCR离线版](https://github.com/PantsuDango/DangoOCR) (@ [PantsuDango](https://github.com/PantsuDango))
- 获取OCR识别结果的key-value [paddleOCRCorrectOutputs](https://github.com/yuranusduke/paddleOCRCorrectOutputs) (@ [yuranusduke](https://github.com/yuranusduke))
- 截屏转文字工具 [scr2txt](https://github.com/lstwzd/scr2txt) (@ [lstwzd](https://github.com/lstwzd))
- 身份证复印件识别 [id_card_ocr](https://github.com/baseli/id_card_ocr)(@ [baseli](https://github.com/baseli))
- 能看懂表格图片的数据助手:[Paddle_Table_Image_Reader](https://github.com/thunder95/Paddle_Table_Image_Reader) (@ [thunder95][https://github.com/thunder95])
- 英文视频自动生成字幕 [AI Studio项目](https://aistudio.baidu.com/aistudio/projectdetail/1054614?channelType=0&channel=0)( @ [叶月水狐](https://aistudio.baidu.com/aistudio/personalcenter/thirdview/322052))
### 1.3 代码与文档优化
- 非常感谢 [zhangxin](https://github.com/ZhangXinNan)([Blog](https://blog.csdn.net/sdlypyzq)) 贡献新的可视化方式、添加.gitgnore、处理手动设置PYTHONPATH环境变量的问题([#210](https://github.com/PaddlePaddle/PaddleOCR/pull/210))。
- 非常感谢 [lyl120117](https://github.com/lyl120117) 贡献打印网络结构的代码([#304](https://github.com/PaddlePaddle/PaddleOCR/pull/304))。
- 非常感谢 [BeyondYourself](https://github.com/BeyondYourself) 给PaddleOCR提了很多非常棒的建议,并简化了PaddleOCR的部分代码风格([so many commits)](https://github.com/PaddlePaddle/PaddleOCR/commits?author=BeyondYourself)
- 非常感谢 [Khanh Tran](https://github.com/xxxpsyduck)[Karl Horky](https://github.com/karlhorky) 贡献修改英文文档。
### 1.4 多语言语料
- 非常感谢 [xiangyubo](https://github.com/xiangyubo) 贡献手写中文OCR数据集([#321](https://github.com/PaddlePaddle/PaddleOCR/pull/321))。
- 非常感谢 [Mejans](https://github.com/Mejans) 给PaddleOCR增加新语言奥克西坦语Occitan的字典和语料([#954](https://github.com/PaddlePaddle/PaddleOCR/pull/954))。
## 2. 贡献说明
### 2.1 新增功能类
PaddleOCR非常欢迎社区贡献以PaddleOCR为核心的各种服务、部署实例与软件应用,经过认证的社区贡献会被添加在上述社区贡献表中,为广大开发者增加曝光,也是PaddleOCR的荣耀,其中:
- 项目形式:官方社区认证的项目代码应有良好的规范和结构,同时,还应配备一个详细的README.md,说明项目的使用方法。通过在requirements.txt文件中增加一行 `paddleocr` 可以自动收录到PaddleOCR的usedby中。
- 合入方式:如果是对PaddleOCR现有工具的更新升级,则会合入主repo。如果为PaddleOCR拓展了新功能,请先与官方人员联系,确认项目是否合入主repo,*即使新功能未合入主repo,我们同样也会以社区贡献的方式为您的个人项目增加曝光。*
### 2.2 代码优化
如果您在使用PaddleOCR时遇到了代码bug、功能不符合预期等问题,可以为PaddleOCR贡献您的修改,其中:
- Python代码规范可参考[附录1:Python代码规范](./code_and_doc.md/#附录1)
- 提交代码前请再三确认不会引入新的bug,并在PR中描述优化点。如果该PR解决了某个issue,请在PR中连接到该issue。所有的PR都应该遵守附录3中的[3.2.10 提交代码的一些约定。](./code_and_doc.md/#提交代码的一些约定)
- 请在提交之前参考下方的[附录3:Pull Request说明](./code_and_doc.md/#附录3)。如果您对git的提交流程不熟悉,同样可以参考附录3的3.2节。
**最后请在PR的题目中加上标签`【third-party】` , 在说明中@Evezerest,拥有此标签的PR将会被高优处理**
### 2.3 文档优化
如果您在使用PaddleOCR时遇到了文档表述不清楚、描述缺失、链接失效等问题,可以为PaddleOCR贡献您的修改。文档书写规范请参考[附录2:文档规范](./code_and_doc.md/#附录2)**最后请在PR的题目中加上标签`【third-party】` , 在说明中@Evezerest,拥有此标签的PR将会被高优处理。**
## 3. 更多贡献机会
我们非常鼓励开发者使用PaddleOCR实现自己的想法,同时我们也列出一些经过分析后认为有价值的拓展方向,供大家参考
- 功能类:IOS端侧demo、前后处理工具、针对各种垂类场景的检测识别模型(如手写体、公式)。
- 文档类:PaddleOCR在各种垂类行业的应用案例(可在公众号中推广)。
## 4. 联系我们
PaddleOCR非常欢迎广大开发者在有意向贡献前与我们联系,这样可以大大降低PR过程中的沟通成本。同时,如果您觉得某些想法个人难以实现,我们也可以通过SIG的形式定向为项目招募志同道合的开发者一起共建。通过SIG渠道贡献的项目将会获得深层次的研发支持与运营资源。
我们推荐的贡献流程是:
- 通过在github issue的题目中增加 `【third-party】` 标记,说明遇到的问题(以及解决的思路)或想拓展的功能,等待值班人员回复。例如 `【third-party】为PaddleOCR贡献IOS示例`
- 与我们沟通确认技术方案或bug、优化点准确无误后进行功能新增或相应的修改,代码与文档遵循相关规范。
- PR链接到上述issue,等待review。
## 5. 致谢与后续
- 合入代码之后,首页README末尾新增感谢贡献,默认链接为github名字及主页,如果有需要更换主页,也可以联系我们。
- 新增重要功能类,会在用户群广而告之,享受开源社区荣誉时刻。
- **如果您有基于PaddleOCR的贡献,但未出现在上述列表中,请按照 `4. 联系我们` 的步骤与我们联系。**
# 模型训练
# PP-OCR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)
- [1.配置文件说明](#配置文件)
- [2. 基本概念](#基本概念)
* [2.1 学习率](#学习率)
* [2.2 正则化](#正则化)
* [2.3 评估指标](#评估指标)
- [3. 数据与垂类场景](#数据与垂类场景)
* [3.1 训练数据](#训练数据)
* [3.2 垂类场景](#垂类场景)
* [3.3 自己构建数据集](#自己构建数据集)
* [4. 常见问题](#常见问题)
<a name="配置文件"></a>
## 1. 配置文件说明
PaddleOCR模型使用配置文件管理网络训练、评估的参数。在配置文件中,可以设置组建模型、优化器、损失函数、模型前后处理的参数,PaddleOCR从配置文件中读取到这些参数,进而组建出完整的训练流程,完成模型训练,在需要对模型进行优化的时,可以通过修改配置文件中的参数完成配置,使用简单且方便修改。
完整的配置文件说明可以参考[配置文件](./config.md)
<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。
## 2. 基本概念
模型调优时需要关注以下参数
模型训练过程中需要手动调整一些超参数,帮助模型以最小的代价获得最优指标。不同的数据量可能需要不同的超参,当您希望在自己的数据上finetune或对模型效果调优时,有以下几个参数调整策略可供参考
<a name="学习率"></a>
### 1.1 学习率
### 2.1 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:
......@@ -42,7 +49,7 @@ Piecewise 代表分段常数衰减,在不同的学习阶段指定不同的学
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
### 2.2 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:
......@@ -55,7 +62,7 @@ Optimizer:
```
<a name="评估指标"></a>
### 1.3 评估指标
### 2.3 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。
......@@ -65,10 +72,10 @@ Optimizer:
<a name="数据与垂类场景"></a>
## 2. 数据与垂类场景
## 3. 数据与垂类场景
<a name="训练数据"></a>
### 2.1 训练数据
### 3.1 训练数据
目前开源的模型,数据集和量级如下:
- 检测:
......@@ -83,13 +90,14 @@ Optimizer:
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
### 3.2 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
### 3.3 自己构建数据集
在构建数据集时有几个经验可供参考:
......@@ -107,7 +115,7 @@ PaddleOCR主要聚焦通用OCR,如果有垂类需求,您可以用PaddleOCR+
<a name="常见问题"></a>
## 3. 常见问题
## 4. 常见问题
**Q**:训练CRNN识别时,如何选择合适的网络输入shape?
......@@ -129,3 +137,14 @@ PaddleOCR主要聚焦通用OCR,如果有垂类需求,您可以用PaddleOCR+
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。
***
具体的训练教程可点击下方链接跳转:
\- [文本检测模型训练](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.3/doc/doc_ch/detection.md)
\- [文本识别模型训练](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.3/doc/doc_ch/recognition.md)
\- [文本方向分类器训练](https://github.com/PaddlePaddle/PaddleOCR/blob/release/2.3/doc/doc_ch/angle_class.md)
......@@ -420,3 +420,5 @@ im_show.save('result.jpg')
| cls | 前向时是否启动分类 (命令行模式下使用use_angle_cls控制前向是否启动分类) | FALSE |
| show_log | 是否打印det和rec等信息 | FALSE |
| type | 执行ocr或者表格结构化, 值可选['ocr','structure'] | ocr |
| ocr_version | OCR模型版本,可选PP-OCRv2, PP-OCR。PP-OCRv2 目前仅支持中文的检测和识别模型,PP-OCR支持中文的检测,识别,多语种识别,方向分类器等模型 | PP-OCRv2 |
| structure_version | 表格结构化模型版本,可选 STRUCTURE。STRUCTURE支持表格结构化模型 | STRUCTURE |
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
此差异已折叠。
Markdown is supported
0% .
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