PaddleOCR Versions Save

Awesome multilingual OCR toolkits based on PaddlePaddle (practical ultra lightweight OCR system, support 80+ languages recognition, provide data annotation and synthesis tools, support training and deployment among server, mobile, embedded and IoT devices)

v2.7.1

5 months ago

New Projects

  • Add Parseq recognition model.(#10836)
  • Add text recognition function to return single character coordinates.(#10515)

New Features

  • Add savefile option to save OCR output results.(#10628)
  • Add more data preprocessing options to ppocr.py.(#10217)
  • A single damaged image does not affect data set inference. (#10216)
  • Compatible with fitz version. (#10181)
  • Compatible with Pillow10.0 upgrade. (#10405)
  • Add Finnish dictionary file. (#10850)
  • Onnxruntime supports GPU. (#10668)
  • TIPC supports XPU and NPU. (#10658, #10460)
  • Add inference on mlu devices. (#10249)

BugFix

  • Fixed cannot find the library error when packaged into exe on windows. (#10502)
  • Fixed the bug of recognize page is affected by the maximum number of PDF files when recognize multiple PDF files. (#10290)
  • Fixed the problem of PPOCRLabel startup failure caused by mismatch in channel number. (#10847)
  • Fix memory leak problem of cpp inference. (#10441)
  • Modify DSR error caused by data enhancement. (#10662)
  • Fix training seed problem. (#10614)
  • Fix table_master tipc error. (#10514)
  • Fixed the problem of error reporting when ppocr.py uses wandb. (#10251)
  • Fix memory leak in predict_rec.py. (#10688)
  • Fixed the issue where dis and iou cannot be calculated correctly due to the index error of structure_boxes in the PaddleStructure::rebuild_table function. (#10810)
  • Compatible with paddle 2.5 fluid exit. (#10391)
  • Fix the performance problem of Tensor.numpy under stride. (#10773)
  • Adapt the size of ABINet during export to the size of ABINetRecResizeImg. (#10892)
  • ABINet training error. (#10744)
  • Fix KeyError in ch_PP-OCRv4_rec_distill.yml. (#10761)

Documentations Fix

  • Fix algorithm_kie_vi_layoutxlm_en.md, kie_datasets_en.md, README.md, algorithm_overview.md, how_to_do_kie_en.md document issues. (#10717)
  • Update documentation issues with setup.py. (#10749)
  • Add pyyaml ​​library in requirements.txt. (#10653)

New Projects

  • 增加Parseq 识别模型。(#10836)
  • 增加文字识别返回单字识别坐标功能。(#10515)

New Features

  • 增加savefile选项,保存ocr输出结果。(#10628)
  • 增加ppocr.py 更多数据预处理选项。(#10217)
  • 单张破损图片不影响整体数据集推理。(#10216)
  • 兼容fitz版本。(#10181)
  • 兼容Pillow10.0升级。(#10405)
  • 增加芬兰语字典文件。(#10850)
  • Onnxruntime 支持GPU。(#10668)
  • TIPC支持XPU、NPU。(#10658,#10460)
  • 增加在mlu设备上的推理。(#10249)

BugFix

  • 修复windows打包成exe找不到库的问题。(#10502)
  • 修复多次识别pdf,受第一次页面最大数量影响的bug。(#10290)
  • 修复通道数不匹配造成的PPOCRLabel启动失败问题。(#10847)
  • 修复cpp推理的内存泄漏问题。(#10441)
  • 修改数据增强导致的DSR报错。(#10662)
  • 修复训练seed传递问题。( #10614)
  • 修复 table_master tipc 报错。(#10514)
  • 修复ppocr.py使用wandb报错问题。(#10251)
  • 修复predict_rec.py中的内存泄漏问题。(#10688)
  • 修复PaddleStructure::rebuild_table函数中structure_boxes的索引错误导致dis和iou无法正确计算的问题。(#10810)
  • 兼容paddle 2.5 fluid退场。 (#10391)
  • 修复stride下Tensor.numpy的性能问题。(#10773)
  • ABINet导出时尺寸适应 ABINetRecResizeImg 的尺寸。(#10892)
  • ABINet训练报错问题。(#10744)
  • 解决ch_PP-OCRv4_rec_distill.yml中的keyError报错问题。(#10761)

Documentations Fix

  • 修复 algorithm_kie_vi_layoutxlm_en.md、kie_datasets_en.md 、README.md、algorithm_overview.md、how_to_do_kie_en.md文档问题(#10717)
  • 更新setup.py的文档问题(#10749)
  • 修复requirements.txt中没有pyyaml库的问题(#10653)

New Contributors

Full Changelog: https://github.com/PaddlePaddle/PaddleOCR/compare/v2.7.0...v2.7.1

v2.7.0

6 months ago

Release Note

  • Release PP-OCRv4, support mobile version and server version
    • PP-OCRv4-mobile:When the speed is comparable, the effect of the Chinese scene is improved by 4.5% compared with PP-OCRv3, the English scene is improved by 10%, and the average recognition accuracy of the 80-language multilingual model is increased by more than 8%.
    • PP-OCRv4-server:Release the OCR model with the highest accuracy at present, the detection model accuracy increased by 4.9% in the Chinese and English scenes, and the recognition model accuracy increased by 2% refer quickstart quick use by one line command, At the same time, the whole process of model training, reasoning, and high-performance deployment can also be completed with few code in the General OCR Industry Solution in PaddleX.
  • ReleasePP-ChatOCR, a new scheme for extracting key information of general scenes using PP-OCR model and ERNIE LLM.

v2.6.0

1 year ago

Release Note

  • Release PP-Structurev2,with functions and performance fully upgraded, adapted to Chinese scenes, and new support for Layout Recovery and one line command to convert PDF to Word;
  • Layout Analysis optimization: model storage reduced by 95%, while speed increased by 11 times, and the average CPU time-cost is only 41ms;
  • Table Recognition optimization: 3 optimization strategies are designed, and the model accuracy is improved by 6% under comparable time consumption;
  • Key Information Extraction optimization:a visual-independent model structure is designed, the accuracy of semantic entity recognition is increased by 2.8%, and the accuracy of relation extraction is increased by 9.1%.

v2.5.0

1 year ago

Release Note

  • Release PP-OCRv3: With comparable speed, the effect of Chinese scene is further improved by 5% compared with PP-OCRv2, the effect of English scene is improved by 11%, and the average recognition accuracy of 80 language multilingual models is improved by more than 5%.
  • Release PPOCRLabelv2: Add the annotation function for table recognition task, key information extraction task and irregular text image.
  • Release interactive e-book "Dive into OCR", covers the cutting-edge theory and code practice of OCR full stack technology.

v2.1.1

2 years ago

Release Note

  1. Newly release model pruning and model quantization tools based on PaddleSlim. Path
  2. Newly release mobile deployment tools based on Paddle-Lite. Path
  3. Newly release Android demo of ppocr system. path
  4. Newly release service deployment based on Paddle Serving. path

v2.1.0

2 years ago

Release Note

  1. Newly release end-to-end text recognition algorithm PGNet which is published in AAAI 2021. Find tutorial here.
  2. Newly release multi language recognition model, support more than 80 languages recognition. Find tutorial here.
  3. Optimize the performance of English recognition model.

v2.0.0

3 years ago

Release Note

一、Support dynamic graph programming paradigm, adapted to Paddle 2.0, including:

  1. Detection algorithm: DB, EAST, SAST
  2. Recognition algorithm: Rosetta, CRNN, RARE, SRN, STAR-Net
  3. PPOCR Chinese models: (1) Detection models: mobile, server (2) Text direction classification models: mobile (3) Recognition models: mobile, server
  4. Multilingual models: (1) English: mobile (2) Japanese, Korean, French, German, etc. 25 languages in total: mobile

二、The related works on deployment have been well adapted, including Inference(Python, C++) , whl, and serving

三、Release the annotation and synthesis tools:

  1. Release a new data synthesis tool, i.e., Style-Text,easy to synthesize a large number of images which are similar to the target scene image.
  2. Release a new data annotation tool, i.e., PPOCRLabel, which is helpful to improve the labeling efficiency. Moreover, the labeling results can be used in training of the PP-OCR system directly.

v1.1.0

3 years ago

3.5M practical ultra lightweight OCR system, support training and deployment among server, mobile, embedded and IoT devices