2020-05-14 11:05:33 +08:00
## 简介
2020-05-14 00:01:00 +08:00
PaddleOCR旨在打造一套丰富、领先、且实用的OCR工具库, 助力使用者训练出更好的模型, 并应用落地。
2020-05-30 11:57:49 +08:00
**近期更新**
- 2020.5.30, 模型预测、训练支持Windows系统, 识别结果的显示进行了优化
- 2020.5.30, 开源通用中文OCR模型
2020-05-30 12:00:49 +08:00
- 2020.5.30, 提供超轻量级中文OCR在线体验
2020-05-30 11:57:49 +08:00
2020-05-14 11:05:33 +08:00
## 特性
2020-05-14 11:12:21 +08:00
- 超轻量级中文OCR, 总模型仅8.6M
- 单模型支持中英文数字组合识别、竖排文本识别、长文本识别
- 检测模型DB( 4.1M) +识别模型CRNN( 4.5M)
2020-05-14 11:05:33 +08:00
- 多种文本检测训练算法, EAST、DB
- 多种文本识别训练算法, Rosetta、CRNN、STAR-Net、RARE
2020-05-30 11:57:49 +08:00
### 支持的中文模型列表:
|模型名称|模型简介|检测模型地址|识别模型地址|
|-|-|-|-|
|chinese_db_crnn_mobile|超轻量级中文OCR模型|[inference模型](https://paddleocr.bj.bcebos.com/ch_models/ch_det_mv3_db_infer.tar) & [预训练模型 ](https://paddleocr.bj.bcebos.com/ch_models/ch_det_mv3_db.tar )|[inference模型](https://paddleocr.bj.bcebos.com/ch_models/ch_rec_mv3_crnn_infer.tar) & [预训练模型 ](https://paddleocr.bj.bcebos.com/ch_models/ch_rec_mv3_crnn.tar )|
|chinese_db_crnn_server|通用中文OCR模型|[inference模型](https://paddleocr.bj.bcebos.com/ch_models/ch_det_r50_vd_db_infer.tar) & [预训练模型 ](https://paddleocr.bj.bcebos.com/ch_models/ch_det_r50_vd_db.tar )|[inference模型](https://paddleocr.bj.bcebos.com/ch_models/ch_rec_r34_vd_crnn_infer.tar) & [预训练模型 ](https://paddleocr.bj.bcebos.com/ch_models/ch_rec_r34_vd_crnn.tar )|
2020-05-29 17:34:56 +08:00
2020-05-30 11:57:49 +08:00
超轻量级中文OCR在线体验地址: https://www.paddlepaddle.org.cn/hub/scene/ocr
2020-05-29 17:34:56 +08:00
2020-05-30 11:57:49 +08:00
**也可以按如下教程快速体验超轻量级中文OCR和通用中文OCR模型。**
2020-05-29 17:34:56 +08:00
2020-05-30 11:57:49 +08:00
## **超轻量级中文OCR以及通用中文OCR体验**
2020-05-14 00:01:00 +08:00
2020-05-14 12:08:11 +08:00
![](doc/imgs_results/11.jpg)
2020-05-14 01:51:17 +08:00
2020-05-14 11:17:08 +08:00
上图是超轻量级中文OCR模型效果展示, 更多效果图请见文末[效果展示](#效果展示)。
2020-05-14 11:05:33 +08:00
2020-05-14 11:20:03 +08:00
#### 1.环境配置
2020-05-14 01:42:35 +08:00
2020-05-14 11:20:03 +08:00
请先参考[快速安装](./doc/installation.md)配置PaddleOCR运行环境。
2020-05-14 00:01:00 +08:00
2020-05-30 11:57:49 +08:00
#### 2.inference模型下载
2020-05-14 01:51:17 +08:00
2020-05-30 11:57:49 +08:00
#### (1)超轻量级中文OCR模型下载
2020-05-14 00:01:00 +08:00
```
2020-05-29 17:34:56 +08:00
mkdir inference & & cd inference
2020-05-30 11:57:49 +08:00
# 下载超轻量级中文OCR模型的检测模型并解压
2020-05-29 17:34:56 +08:00
wget https://paddleocr.bj.bcebos.com/ch_models/ch_det_mv3_db_infer.tar & & tar xf ch_det_mv3_db_infer.tar
2020-05-30 11:57:49 +08:00
# 下载超轻量级中文OCR模型的识别模型并解压
2020-05-29 17:34:56 +08:00
wget https://paddleocr.bj.bcebos.com/ch_models/ch_rec_mv3_crnn_infer.tar & & tar xf ch_rec_mv3_crnn_infer.tar
2020-05-30 11:57:49 +08:00
cd ..
```
#### (2)通用中文OCR模型下载
```
mkdir inference & & cd inference
# 下载通用中文OCR模型的检测模型并解压
wget https://paddleocr.bj.bcebos.com/ch_models/ch_det_r50_vd_db_infer.tar & & tar xf ch_det_r50_vd_db_infer.tar
# 下载通用中文OCR模型的识别模型并解压
wget https://paddleocr.bj.bcebos.com/ch_models/ch_rec_r34_vd_crnn_infer.tar & & tar xf ch_rec_r34_vd_crnn_infer.tar
cd ..
2020-05-14 00:01:00 +08:00
```
2020-05-14 11:05:33 +08:00
#### 3.单张图像或者图像集合预测
2020-05-14 14:29:03 +08:00
以下代码实现了文本检测、识别串联推理, 在执行预测时, 需要通过参数image_dir指定单张图像或者图像集合的路径、参数det_model_dir指定检测inference模型的路径和参数rec_model_dir指定识别inference模型的路径。可视化识别结果默认保存到 ./inference_results 文件夹里面。
2020-05-14 11:05:33 +08:00
2020-05-14 00:01:00 +08:00
```
2020-05-14 11:05:33 +08:00
# 设置PYTHONPATH环境变量
2020-05-14 00:01:00 +08:00
export PYTHONPATH=.
2020-05-14 11:05:33 +08:00
# 预测image_dir指定的单张图像
2020-05-30 11:57:49 +08:00
python3 tools/infer/predict_system.py --image_dir="./doc/imgs/11.jpg" --det_model_dir="./inference/ch_det_mv3_db/" --rec_model_dir="./inference/ch_rec_mv3_crnn/"
2020-05-14 11:05:33 +08:00
# 预测image_dir指定的图像集合
2020-05-30 11:57:49 +08:00
python3 tools/infer/predict_system.py --image_dir="./doc/imgs/" --det_model_dir="./inference/ch_det_mv3_db/" --rec_model_dir="./inference/ch_rec_mv3_crnn/"
2020-05-15 22:07:18 +08:00
2020-05-29 17:34:56 +08:00
# 如果想使用CPU进行预测, 需设置use_gpu参数为False
2020-05-30 11:57:49 +08:00
python3 tools/infer/predict_system.py --image_dir="./doc/imgs/11.jpg" --det_model_dir="./inference/ch_det_mv3_db/" --rec_model_dir="./inference/ch_rec_mv3_crnn/" --use_gpu=False
```
通用中文OCR模型的体验可以按照上述步骤下载相应的模型, 并且更新相关的参数, 示例如下:
```
# 预测image_dir指定的单张图像
python3 tools/infer/predict_system.py --image_dir="./doc/imgs/11.jpg" --det_model_dir="./inference/ch_det_r50_vd_db/" --rec_model_dir="./inference/ch_rec_r34_vd_crnn/"
2020-05-14 00:01:00 +08:00
```
2020-05-29 17:34:56 +08:00
2020-05-14 14:55:39 +08:00
更多的文本检测、识别串联推理使用方式请参考文档教程中[基于预测引擎推理](./doc/inference.md)。
2020-05-14 00:01:00 +08:00
2020-05-14 01:40:20 +08:00
## 文档教程
- [快速安装 ](./doc/installation.md )
2020-05-29 17:36:55 +08:00
- [文本检测模型训练/评估/预测 ](./doc/detection.md )
- [文本识别模型训练/评估/预测 ](./doc/recognition.md )
- [基于预测引擎推理 ](./doc/inference.md )
2020-05-14 01:40:20 +08:00
2020-05-14 11:05:33 +08:00
## 文本检测算法
2020-05-14 00:01:00 +08:00
PaddleOCR开源的文本检测算法列表:
2020-05-14 12:04:43 +08:00
- [x] EAST([paper](https://arxiv.org/abs/1704.03155))
2020-05-14 12:07:22 +08:00
- [x] DB([paper](https://arxiv.org/abs/1911.08947))
- [ ] SAST([paper](https://arxiv.org/abs/1908.05498))(百度自研, comming soon)
2020-05-14 00:01:00 +08:00
2020-05-14 11:05:33 +08:00
在ICDAR2015文本检测公开数据集上, 算法效果如下:
2020-05-14 00:01:00 +08:00
2020-05-25 16:50:07 +08:00
|模型|骨干网络|precision|recall|Hmean|下载链接|
2020-05-25 18:14:13 +08:00
|-|-|-|-|-|-|
2020-05-29 16:43:54 +08:00
|EAST|ResNet50_vd|88.18%|85.51%|86.82%|[下载链接](https://paddleocr.bj.bcebos.com/det_r50_vd_east.tar)|
2020-05-25 16:50:07 +08:00
|EAST|MobileNetV3|81.67%|79.83%|80.74%|[下载链接](https://paddleocr.bj.bcebos.com/det_mv3_east.tar)|
|DB|ResNet50_vd|83.79%|80.65%|82.19%|[下载链接](https://paddleocr.bj.bcebos.com/det_r50_vd_db.tar)|
|DB|MobileNetV3|75.92%|73.18%|74.53%|[下载链接](https://paddleocr.bj.bcebos.com/det_mv3_db.tar)|
2020-05-25 16:29:20 +08:00
2020-05-25 18:14:13 +08:00
* 注: 上述DB模型的训练和评估, 需设置后处理参数box_thresh=0.6, unclip_ratio=1.5,使用不同数据集、不同模型训练,可调整这两个参数进行优化
2020-05-14 00:01:00 +08:00
2020-05-14 11:05:33 +08:00
PaddleOCR文本检测算法的训练和使用请参考文档教程中[文本检测模型训练/评估/预测](./doc/detection.md)。
2020-05-14 00:01:00 +08:00
2020-05-14 11:05:33 +08:00
## 文本识别算法
2020-05-14 00:01:00 +08:00
PaddleOCR开源的文本识别算法列表:
2020-05-14 12:04:43 +08:00
- [x] CRNN([paper](https://arxiv.org/abs/1507.05717))
- [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))
- [ ] SRN([paper](https://arxiv.org/abs/2003.12294))(百度自研, comming soon)
2020-05-14 00:01:00 +08:00
2020-05-14 11:30:40 +08:00
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:
2020-05-14 00:01:00 +08:00
2020-05-15 19:51:25 +08:00
|模型|骨干网络|Avg Accuracy|模型存储命名|下载链接|
2020-05-15 19:51:49 +08:00
|-|-|-|-|-|
2020-05-15 19:51:25 +08:00
|Rosetta|Resnet34_vd|80.24%|rec_r34_vd_none_none_ctc|[下载链接](https://paddleocr.bj.bcebos.com/rec_r34_vd_none_none_ctc.tar)|
|Rosetta|MobileNetV3|78.16%|rec_mv3_none_none_ctc|[下载链接](https://paddleocr.bj.bcebos.com/rec_mv3_none_none_ctc.tar)|
|CRNN|Resnet34_vd|82.20%|rec_r34_vd_none_bilstm_ctc|[下载链接](https://paddleocr.bj.bcebos.com/rec_r34_vd_none_bilstm_ctc.tar)|
|CRNN|MobileNetV3|79.37%|rec_mv3_none_bilstm_ctc|[下载链接](https://paddleocr.bj.bcebos.com/rec_mv3_none_bilstm_ctc.tar)|
|STAR-Net|Resnet34_vd|83.93%|rec_r34_vd_tps_bilstm_ctc|[下载链接](https://paddleocr.bj.bcebos.com/rec_r34_vd_tps_bilstm_ctc.tar)|
|STAR-Net|MobileNetV3|81.56%|rec_mv3_tps_bilstm_ctc|[下载链接](https://paddleocr.bj.bcebos.com/rec_mv3_tps_bilstm_ctc.tar)|
|RARE|Resnet34_vd|84.90%|rec_r34_vd_tps_bilstm_attn|[下载链接](https://paddleocr.bj.bcebos.com/rec_r34_vd_tps_bilstm_attn.tar)|
|RARE|MobileNetV3|83.32%|rec_mv3_tps_bilstm_attn|[下载链接](https://paddleocr.bj.bcebos.com/rec_mv3_tps_bilstm_attn.tar)|
2020-05-14 00:01:00 +08:00
2020-05-14 11:05:33 +08:00
PaddleOCR文本识别算法的训练和使用请参考文档教程中[文本识别模型训练/评估/预测](./doc/recognition.md)。
2020-05-14 00:01:00 +08:00
2020-05-14 11:05:33 +08:00
## 端到端OCR算法
- [ ] [End2End-PSL ](https://arxiv.org/abs/1909.07808 )(百度自研, comming soon)
2020-05-14 00:01:00 +08:00
2020-05-14 11:16:40 +08:00
< a name = "效果展示" > < / a >
2020-05-30 12:06:31 +08:00
## 超轻量级中文OCR效果展示
2020-05-14 12:08:11 +08:00
![](doc/imgs_results/1.jpg)
![](doc/imgs_results/7.jpg)
![](doc/imgs_results/12.jpg)
![](doc/imgs_results/4.jpg)
![](doc/imgs_results/6.jpg)
![](doc/imgs_results/9.jpg)
![](doc/imgs_results/16.png)
![](doc/imgs_results/22.jpg)
2020-05-14 00:01:00 +08:00
2020-06-02 16:47:04 +08:00
## 通用中文OCR效果展示
![](doc/imgs_results/chinese_db_crnn_server/11.jpg)
![](doc/imgs_results/chinese_db_crnn_server/2.jpg)
![](doc/imgs_results/chinese_db_crnn_server/8.jpg)
2020-06-02 16:23:45 +08:00
## 更新
- 2020.5.30, 模型预测、训练支持Windows系统, 识别结果的显示进行了优化
- 2020.5.30, 开源通用中文OCR模型
- 2020.5.30, 提供超轻量级中文OCR在线体验
2020-05-14 00:01:00 +08:00
2020-05-14 01:58:27 +08:00
## 参考文献
2020-05-14 00:01:00 +08:00
```
1. EAST:
@inproceedings {zhou2017east,
title={EAST: an efficient and accurate scene text detector},
author={Zhou, Xinyu and Yao, Cong and Wen, He and Wang, Yuzhi and Zhou, Shuchang and He, Weiran and Liang, Jiajun},
booktitle={Proceedings of the IEEE conference on Computer Vision and Pattern Recognition},
pages={5551--5560},
year={2017}
}
2. DB:
@article {liao2019real,
title={Real-time Scene Text Detection with Differentiable Binarization},
author={Liao, Minghui and Wan, Zhaoyi and Yao, Cong and Chen, Kai and Bai, Xiang},
journal={arXiv preprint arXiv:1911.08947},
year={2019}
}
3. DTRB:
@inproceedings {baek2019wrong,
title={What is wrong with scene text recognition model comparisons? dataset and model analysis},
author={Baek, Jeonghun and Kim, Geewook and Lee, Junyeop and Park, Sungrae and Han, Dongyoon and Yun, Sangdoo and Oh, Seong Joon and Lee, Hwalsuk},
booktitle={Proceedings of the IEEE International Conference on Computer Vision},
pages={4715--4723},
year={2019}
}
4. SAST:
@inproceedings {wang2019single,
title={A Single-Shot Arbitrarily-Shaped Text Detector based on Context Attended Multi-Task Learning},
author={Wang, Pengfei and Zhang, Chengquan and Qi, Fei and Huang, Zuming and En, Mengyi and Han, Junyu and Liu, Jingtuo and Ding, Errui and Shi, Guangming},
booktitle={Proceedings of the 27th ACM International Conference on Multimedia},
pages={1277--1285},
year={2019}
}
5. SRN:
@article {yu2020towards,
title={Towards Accurate Scene Text Recognition with Semantic Reasoning Networks},
author={Yu, Deli and Li, Xuan and Zhang, Chengquan and Han, Junyu and Liu, Jingtuo and Ding, Errui},
journal={arXiv preprint arXiv:2003.12294},
year={2020}
}
6. end2end-psl:
@inproceedings {sun2019chinese,
title={Chinese Street View Text: Large-scale Chinese Text Reading with Partially Supervised Learning},
author={Sun, Yipeng and Liu, Jiaming and Liu, Wei and Han, Junyu and Ding, Errui and Liu, Jingtuo},
booktitle={Proceedings of the IEEE International Conference on Computer Vision},
pages={9086--9095},
year={2019}
}
```
2020-05-14 01:58:00 +08:00
## 许可证书
本项目的发布受< a href = "https://github.com/PaddlePaddle/PaddleOCR/blob/master/LICENSE" > Apache 2.0 license< / a > 许可认证。
## 如何贡献代码
我们非常欢迎你为PaddleOCR贡献代码, 也十分感谢你的反馈。