Starstruk Logo MarkStarstruk

v2.8.0

PaddlePaddle/PaddleOCRv2.8.0Jul 4, 2024by GreatV

AI Summary

A major release introducing PP-OCRv4 and several new algorithm models (ParseQ, SVTR Large, CPPD). It focuses on performance, multi-language support, and ONNX runtime improvements.

Key Highlights

  • Released PP-OCRv4 (mobile and server) with significant accuracy improvements.
  • Added ParseQ recognition method.
  • Added SVTR Large model.
  • Added CPPD (Compact Prediction Dense) model.
  • Added ONNX Runtime GPU support.

Breaking Changes

  • Removed `fluid` API (Paddle 2.6+ requirement).
  • Temporarily set numpy < 2.0.

New Features

  • PP-OCRv4 models
  • ParseQ model
  • SVTR Large model
  • CPPD model
  • ONNX Runtime GPU support
  • XPU/NPU support
  • Burmese language support

Full Release Notes

**[终于等到你！PaddleOCR 新版本发布！](https://pfcc.blog/posts/paddleocr-release)**

## What's Changed
* [Cherry-pick] #10515 by @ToddBear in https://github.com/PaddlePaddle/PaddleOCR/pull/10537
* [BugFix]compat_pillow by @shiyutang in https://github.com/PaddlePaddle/PaddleOCR/pull/10596
* [bug fix] fix none res in recovery by @andyjiang1116 in https://github.com/PaddlePaddle/PaddleOCR/pull/10603
* Fix seed passing issue of build_dataloader by @RuohengMa in https://github.com/PaddlePaddle/PaddleOCR/pull/10614
* [bug fix]rm invalid params by @andyjiang1116 in https://github.com/PaddlePaddle/PaddleOCR/pull/10605
* [Cherry-pick] #10441 #10512 by @moehuster in https://github.com/PaddlePaddle/PaddleOCR/pull/10593
* 修改数据增强导致的DSR报错 by @xu-peng-7 in https://github.com/PaddlePaddle/PaddleOCR/pull/10662
* onnxruntime support gpu by @WenmuZhou in https://github.com/PaddlePaddle/PaddleOCR/pull/10668
* Update VQA to use the updated LayoutLM syntax from PaddleNLP by @sijunhe in https://github.com/PaddlePaddle/PaddleOCR/pull/9791
* 实现功能：当--savefile为true时，在--output下以当前图片名称后接“.txt”为文件名保存ocr推理结果，解决了issues： by @WilliamQf-AI in https://github.com/PaddlePaddle/PaddleOCR/pull/10628
* Cherrypicking GH-10217 and GH-10216 to PaddlePaddle:dygraph by @UserUnknownFactor in https://github.com/PaddlePaddle/PaddleOCR/pull/10654
* fix numpy speed by @wanghuancoder in https://github.com/PaddlePaddle/PaddleOCR/pull/10773
* Cherrypicking GH-10251 & GH-10181 to PaddleOCR:dygraph by @itasli in https://github.com/PaddlePaddle/PaddleOCR/pull/10710
* rec_r45_abinet.yml add max_length and image_size by @xlg-go in https://github.com/PaddlePaddle/PaddleOCR/pull/10744
* ch_PP-OCRv4_rec_distill.yml, fix KeyError: 'NRTRLabelDecode' by @xlg-go in https://github.com/PaddlePaddle/PaddleOCR/pull/10761
* 根据推理对三通道的图像需求，以及opencv中imread参数说明IMREAD_COLOR(If set, always convert … by @Gmgge in https://github.com/PaddlePaddle/PaddleOCR/pull/10777
* Update algorithm_kie_vi_layoutxlm_en.md by @sagarjgb in https://github.com/PaddlePaddle/PaddleOCR/pull/10736
* Add new recognition method "ParseQ" by @ToddBear in https://github.com/PaddlePaddle/PaddleOCR/pull/10836
* rm fluid for paddle dev by @tink2123 in https://github.com/PaddlePaddle/PaddleOCR/pull/10931
* rec_r45_abinet for export model by @xlg-go in https://github.com/PaddlePaddle/PaddleOCR/pull/10892
* fix:修复通道数不匹配造成的PPOCRLabel启动失败问题#10748,根据更新日志发现#10655，由于paddleocr中增加了对… by @Gmgge in https://github.com/PaddlePaddle/PaddleOCR/pull/10847
* [New] add rec CPPD model by @Topdu in https://github.com/PaddlePaddle/PaddleOCR/pull/10990
* fix `cls_x` and `bbox_x` is possibly unbound by @SigureMo in https://github.com/PaddlePaddle/PaddleOCR/pull/10991
* add svtr large model by @zhangyubo0722 in https://github.com/PaddlePaddle/PaddleOCR/pull/10937
* [WIP]support eval pre epoch by @zhangyubo0722 in https://github.com/PaddlePaddle/PaddleOCR/pull/11003
* Update kie_datasets_en.md by @sagarjgb in https://github.com/PaddlePaddle/PaddleOCR/pull/10735
* fix import collection for py310 by @tink2123 in https://github.com/PaddlePaddle/PaddleOCR/pull/11012
* update ppocrv4_framework by @tink2123 in https://github.com/PaddlePaddle/PaddleOCR/pull/11048
* Update how_to_do_kie_en.md by @sagarjgb in https://github.com/PaddlePaddle/PaddleOCR/pull/10731
* add cppd u14m train model and doc by @Topdu in https://github.com/PaddlePaddle/PaddleOCR/pull/11052
* Fixed bug with "max_text_length" for VisionLAN by @victor30608 in https://github.com/PaddlePaddle/PaddleOCR/pull/11025
* Cherrypicking GH-10923 to PaddleOCR:dygraph by @itasli in https://github.com/PaddlePaddle/PaddleOCR/pull/11069
* Update quickstart_en.md by @sagarjgb in https://github.com/PaddlePaddle/PaddleOCR/pull/10732
* Update README.md by @sagarjgb in https://github.com/PaddlePaddle/PaddleOCR/pull/10733
* Update algorithm_overview_en.md by @sagarjgb in https://github.com/PaddlePaddle/PaddleOCR/pull/10734
* [Cherry-pick] Cherry-pick from release/2.6 by @shiyutang in https://github.com/PaddlePaddle/PaddleOCR/pull/11092
* [TIPC]update tipc scripts by @USTCKAY in https://github.com/PaddlePaddle/PaddleOCR/pull/11097
* fix satrn export for paddle2.5 by @tink2123 in https://github.com/PaddlePaddle/PaddleOCR/pull/11096
* [BugFix]Fix parseq net by @shiyutang in https://github.com/PaddlePaddle/PaddleOCR/pull/11126
* update uygur dict by @hfengzhi in https://github.com/PaddlePaddle/PaddleOCR/pull/11125
* Add tipc for "ParseQ" method by @ToddBear in https://github.com/PaddlePaddle/PaddleOCR/pull/10843
* fix SAR inference, when batch size>1, norm_img_batch and valid_ratios… by @shiyunalex in https://github.com/PaddlePaddle/PaddleOCR/pull/11238
* v4 det cml configs by @sylarwcy in https://github.com/PaddlePaddle/PaddleOCR/pull/11258
* 解決recognition的train test分割程式執行後的文檔每行間多出一行空格 by @DingHsun in https://github.com/PaddlePaddle/PaddleOCR/pull/11280
* Fix for Ambiguous Boolean Evaluation Error in PaddleOCR with Python 3.11 by @muhammadAgfian96 in https://github.com/PaddlePaddle/PaddleOCR/pull/11287
* Dygraph【benchmark】add max_mem_reserved for benchmark by @mmglove in https://github.com/PaddlePaddle/PaddleOCR/pull/11284
* Fix bug when running on XPU by @RuohengMa in https://github.com/PaddlePaddle/PaddleOCR/pull/11299
* Dygraph by @RuohengMa in https://github.com/PaddlePaddle/PaddleOCR/pull/11301
* Dygraph fix max_mem_reserved for benchmark by @mmglove in https://github.com/PaddlePaddle/PaddleOCR/pull/11341
* 在check_gpu时增加对当前环境可用设备的检查 by @TracebaK in https://github.com/PaddlePaddle/PaddleOCR/pull/11293
* Fixed some bugs that caused PPOCRLabel to crash, added ability to expand checkboxes by @g39088902 in https://github.com/PaddlePaddle/PaddleOCR/pull/11236
* fix a bug for rec_postprocess.py by @Ataraxy33 in https://github.com/PaddlePaddle/PaddleOCR/pull/11389
* Optimize prediction on long image and deduplicate similar boxes with multiple lables by @marswen in https://github.com/PaddlePaddle/PaddleOCR/pull/11366
* doc: add doc for satrn by @wkml in https://github.com/PaddlePaddle/PaddleOCR/pull/11397
* Update zeros' comment in rec_abinet_head.py by @YesianRohn in https://github.com/PaddlePaddle/PaddleOCR/pull/11374
* Fix QPointF IndexError: list index out of range by @firmament2008 in https://github.com/PaddlePaddle/PaddleOCR/pull/11393
* update paddlex of readme by @zhangyubo0722 in https://github.com/PaddlePaddle/PaddleOCR/pull/11422
* chore: add notes for docker gpu deploy PP-OCRv4 by @sheiy in https://github.com/PaddlePaddle/PaddleOCR/pull/11390
* Fix words by @co63oc in https://github.com/PaddlePaddle/PaddleOCR/pull/11448
* [Feature]Complete the ppocrv4_act by @ranchongzhi in https://github.com/PaddlePaddle/PaddleOCR/pull/11345
* rm QR code in the document by @tink2123 in https://github.com/PaddlePaddle/PaddleOCR/pull/11512
* rm QR code by @tink2123 in https://github.com/PaddlePaddle/PaddleOCR/pull/11532
* Fix dead links by @MatKollar in https://github.com/PaddlePaddle/PaddleOCR/pull/11520
* cherry-pick for lazy import pymupdf and pre-commit by @tink2123 in https://github.com/PaddlePaddle/PaddleOCR/pull/11692
* adapter new type promotion rule for Paddle 2.6 by @zxcd in https://github.com/PaddlePaddle/PaddleOCR/pull/11698
* setup a workflow for publishing package to pypi by @jzhang533 in https://github.com/PaddlePaddle/PaddleOCR/pull/11804
* update link mentioned at #11763 by @jzhang533 in https://github.com/PaddlePaddle/PaddleOCR/pull/11764
* fix AttributeError by @GreatV in https://github.com/PaddlePaddle/PaddleOCR/pull/11686
* fix: Correct misuse of `try_import` from `paddle.utils` by @neteroster in https://github.com/PaddlePaddle/PaddleOCR/pull/11820
* Update quickstart.md for a better python pdf demo by @qwedc001 in https://github.com/PaddlePaddle/PaddleOCR/pull/11927
* Update quickstart_en.md by @qwedc001 in https://github.com/PaddlePaddle/PaddleOCR/pull/11934
* Enhance the OCR recognition accuracy of PPStructure. by @RussellLuo in https://github.com/PaddlePaddle/PaddleOCR/pull/11916
* add u14m results of cppd by @Topdu in https://github.com/PaddlePaddle/PaddleOCR/pull/11943
* use tensor.shape bug not paddle.shape(tensor) by @wanghuancoder in https://github.com/PaddlePaddle/PaddleOCR/pull/11919
* add pre-commit workflow by @GreatV in https://github.com/PaddlePaddle/PaddleOCR/pull/11973
* docs: Update FAQ.md, delete repeated question by @xu8117 in https://github.com/PaddlePaddle/PaddleOCR/pull/11972
* Fix the bug where Python scripts fail to execute PDF text recognition… by @guangyunms in https://github.com/PaddlePaddle/PaddleOCR/pull/11994
* 【OCR Issue No.9】以可选形式支持Visualdl by @Liyulingyue in https://github.com/PaddlePaddle/PaddleOCR/pull/11947
* fix weird version info by @GreatV in https://github.com/PaddlePaddle/PaddleOCR/pull/12003
* 【OCR Issue No.9】移除明确不适合放在ppocr依赖中的依赖项 by @Liyulingyue in https://github.com/PaddlePaddle/PaddleOCR/pull/11946
* Burmese Language dict and corpus by @1chimaruGin in https://github.com/PaddlePaddle/PaddleOCR/pull/12020
* 面版识别添加onnx支持完善 by @heweisheng in https://github.com/PaddlePaddle/PaddleOCR/pull/12068
* Update README.md by @dyning in https://github.com/PaddlePaddle/PaddleOCR/pull/12086
* fix readme codestyle by @GreatV in https://github.com/PaddlePaddle/PaddleOCR/pull/12095
* fix wrong link for 通用OCR in README.txt by @tackhwa in https://github.com/PaddlePaddle/PaddleOCR/pull/12100
* move PPOCRLabel to PFCCLab/PPOCRLabel by @GreatV in https://github.com/PaddlePaddle/PaddleOCR/pull/12104
* move StyleText to PFCCLab/StyleText by @GreatV in https://github.com/PaddlePaddle/PaddleOCR/pull/12121
* openocr compti code by @Topdu in https://github.com/PaddlePaddle/PaddleOCR/pull/12033
* table rec code by @invictuszhao in https://github.com/PaddlePaddle/PaddleOCR/pull/11999
* Error with pyclipper inhomogeneous expanded array by @zovelsanj in https://github.com/PaddlePaddle/PaddleOCR/pull/12108
* 【OCR Issue No.2】修复训练过程中找不到對應模型和训练时计算精度报错 by @mattheliu in https://github.com/PaddlePaddle/PaddleOCR/pull/12140
* fix the issue of repeatedly downloading pretrained model by @GreatV in https://github.com/PaddlePaddle/PaddleOCR/pull/12142
* fix typo by @GreatV in https://github.com/PaddlePaddle/PaddleOCR/pull/12146
* Added Bengali , gujrati and kazakh dictionary by @asif-ca in https://github.com/PaddlePaddle/PaddleOCR/pull/12151
* add ci for paddleocr test by @GreatV in https://github.com/PaddlePaddle/PaddleOCR/pull/12062
* update community section of README, and did a few tweaks by @jzhang533 in https://github.com/PaddlePaddle/PaddleOCR/pull/12154
* 【OCR Issue No.12】Modify the setuptools configuration from SETUP.py into PYPROJECT.toml by @Liyulingyue in https://github.com/PaddlePaddle/PaddleOCR/pull/12013
* commit fix by running pre-commit run -a by @jzhang533 in https://github.com/PaddlePaddle/PaddleOCR/pull/12165
* added sliding window for large image inference by @aspaul20 in https://github.com/PaddlePaddle/PaddleOCR/pull/12152
* chore(issue_template): Optimize issue template and add Dissucussion entry by @SWHL in https://github.com/PaddlePaddle/PaddleOCR/pull/12171
* fix the error of `np.int` by @Liyulingyue in https://github.com/PaddlePaddle/PaddleOCR/pull/12249
* add `slice` op demo for quickstart by @GreatV in https://github.com/PaddlePaddle/PaddleOCR/pull/12439
* fix and enable flake8 E721 by @GreatV in https://github.com/PaddlePaddle/PaddleOCR/pull/12258
* Update pyproject.toml by @GreatV in https://github.com/PaddlePaddle/PaddleOCR/pull/12520
* Save detection inference logs to file by @AlexPasqua in https://github.com/PaddlePaddle/PaddleOCR/pull/12042
* update common pre-commit configs by @jzhang533 in https://github.com/PaddlePaddle/PaddleOCR/pull/12516
* Update FAQ.md by @SkyFlap in https://github.com/PaddlePaddle/PaddleOCR/pull/12528
* Save recognition inference logs to file by @AlexPasqua in https://github.com/PaddlePaddle/PaddleOCR/pull/12542
* add test for cls_postprocess by @GreatV in https://github.com/PaddlePaddle/PaddleOCR/pull/12534
* fix layout recovery error: list index out of range by @GreatV in https://github.com/PaddlePaddle/PaddleOCR/pull/12541
* update documentation about the installation by @GreatV in https://github.com/PaddlePaddle/PaddleOCR/pull/12565
* Update README.md by @SkyFlap in https://github.com/PaddlePaddle/PaddleOCR/pull/12583
* fix wrong output when --det=False by @GreatV in https://github.com/PaddlePaddle/PaddleOCR/pull/12589
* fix requirements.txt by @wencan in https://github.com/PaddlePaddle/PaddleOCR/pull/12605
* Add note about issue and Discussions on README by @SWHL in https://github.com/PaddlePaddle/PaddleOCR/pull/12703
* 【Doc】Add notification document by @tink2123 in https://github.com/PaddlePaddle/PaddleOCR/pull/12731
* docs(README): Add Discord shields by @SWHL in https://github.com/PaddlePaddle/PaddleOCR/pull/12755
* [doc] Fix wrong link by @enximi in https://github.com/PaddlePaddle/PaddleOCR/pull/12824
* fix wrong function name by @enximi in https://github.com/PaddlePaddle/PaddleOCR/pull/12826
* [doc] fixing broken links and updating paddle2onnx usage descriptions by @Mrxuba in https://github.com/PaddlePaddle/PaddleOCR/pull/12749
* add layout score return by @myhloli in https://github.com/PaddlePaddle/PaddleOCR/pull/13068
* fix pr11999 about slanet train error. by @UserWangZz in https://github.com/PaddlePaddle/PaddleOCR/pull/13073
* temporarily set numpy < 2.0 by @GreatV in https://github.com/PaddlePaddle/PaddleOCR/pull/13136
* docs: Fixed the invalid Discord invite link by @SWHL in https://github.com/PaddlePaddle/PaddleOCR/pull/13135
* fix(sec): upgrade fonttools to 4.43.0 by @chncaption in https://github.com/PaddlePaddle/PaddleOCR/pull/13091
* update ppocrv4 docs by @mattheliu in https://github.com/PaddlePaddle/PaddleOCR/pull/13081
* 解决onnruntime推理问题 by @Ghorc in https://github.com/PaddlePaddle/PaddleOCR/pull/13154
* fix opencv import error for numpy 2.0 by @GreatV in https://github.com/PaddlePaddle/PaddleOCR/pull/13105
* The model link for tabe rec compti by @invictuszhao in https://github.com/PaddlePaddle/PaddleOCR/pull/13254
* Updated Recognition Competition Model Link by @Topdu in https://github.com/PaddlePaddle/PaddleOCR/pull/13259
* update ReadMe by @GreatV in https://github.com/PaddlePaddle/PaddleOCR/pull/13263

## New Contributors
* @RuohengMa made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/10614
* @sijunhe made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/9791
* @WilliamQf-AI made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/10628
* @xlg-go made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/10744
* @Gmgge made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/10777
* @victor30608 made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/11025
* @hfengzhi made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/11125
* @shiyunalex made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/11238
* @sylarwcy made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/11258
* @DingHsun made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/11280
* @muhammadAgfian96 made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/11287
* @TracebaK made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/11293
* @g39088902 made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/11236
* @marswen made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/11366
* @wkml made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/11397
* @YesianRohn made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/11374
* @firmament2008 made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/11393
* @sheiy made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/11390
* @co63oc made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/11448
* @ranchongzhi made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/11345
* @MatKollar made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/11520
* @neteroster made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/11820
* @qwedc001 made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/11927
* @RussellLuo made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/11916
* @xu8117 made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/11972
* @guangyunms made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/11994
* @Liyulingyue made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/11947
* @1chimaruGin made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/12020
* @heweisheng made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/12068
* @tackhwa made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/12100
* @invictuszhao made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/11999
* @zovelsanj made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/12108
* @mattheliu made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/12140
* @asif-ca made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/12151
* @aspaul20 made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/12152
* @SWHL made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/12171
* @AlexPasqua made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/12042
* @SkyFlap made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/12528
* @wencan made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/12605
* @enximi made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/12824
* @Mrxuba made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/12749
* @myhloli made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/13068
* @chncaption made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/13091
* @Ghorc made their first contribution in https://github.com/PaddlePaddle/PaddleOCR/pull/13154

**Full Changelog**: https://github.com/PaddlePaddle/PaddleOCR/compare/v2.7.5...v2.8.0