Overview

7 Pull requests merged by3 people

• update mcp into main readme

  #15953 mergedJul 11, 2025

• [cherry-pick] update legacy feature doc (#16012)

  #16013 mergedJul 9, 2025

• update legacy feature doc

  #16012 mergedJul 9, 2025

• [Cherry-Pick] Replace python-magic with puremagic (#15990)

  #15993 mergedJul 9, 2025

• [Feat][MCP] Replace python-magic with puremagic

  #15990 mergedJul 8, 2025

• [Cherry-pick] update layout for mcp readme demo 3

  #15976 mergedJul 7, 2025

• update layout for demo 3

  #15975 mergedJul 7, 2025

6 Pull requests opened by4 people

• bugfix: params setting in init is not work

  #15997 openedJul 8, 2025

• fix formula dependency

  #16003 openedJul 8, 2025

• [Docs] Fix errors

  #16005 openedJul 9, 2025

• adapt PaddlePaddle/PaddleX/pull/4312

  #16009 openedJul 9, 2025

• [WIP][Feat] Accommodate new PaddlePredictorOption API

  #16017 openedJul 10, 2025

• Add multilingual PP-OCRv5 model support in text recognition inference

  #16032 openedJul 12, 2025

21 Issues closed by10 people

• 在 macOS 上使用 paddleocr 的 PPStructureV3 时报错。似乎是它试图调用一个在当前 paddlepaddle 3.0.0 版本中不存在的方法

  #15629 closedJul 12, 2025

• 0 precision and recall when training the RE model after the SER one

  #16023 closedJul 10, 2025

• Device id must be less than GPU count, but received id is: 3. GPU count is: 1

  #15497 closedJul 10, 2025

• Crashing Google Colab

  #15619 closedJul 10, 2025

• 安装训练依赖时报错：ultra-infer-gpu-python 1.1.1 requires numpy<2, but you have numpy 2.2.6 which is incompatible.

  #16004 closedJul 9, 2025

• AttributeError: 'bool' object has no attribute 'sum' when training with custom dataset

  #16002 closedJul 9, 2025

• multiprocessing.spawn on Linux imports paddleocr.py module instead of paddleocr package, causing ModuleNotFoundError for submodules

  #15867 closedJul 9, 2025

• ppstructure/kie/predict_kie_token_ser 推理时 很多以前的参数不支持

  #15840 closedJul 9, 2025

• Nuitka 打包问题关于paddlex 和paddleocr3.0

  #15767 closedJul 9, 2025

• paddleocr3.0 转换trt 出现算子问题

  #15769 closedJul 9, 2025

• 2.9.1好用，2.10.0 No text found，可能有bug

  #14860 closedJul 9, 2025

• rec_polys得到的坐标框在原图中和rec_text不符

  #15992 closedJul 8, 2025

• 文本检测模型二次微调，导出模型后识别精度下降

  #15974 closedJul 8, 2025

• paddle OCR模型是否可以拿来商业使用？

  #15985 closedJul 8, 2025

• Multi Language support of Paddle OCR3.0

  #15551 closedJul 7, 2025

• paddleocr3.0.0字体问题

  #15607 closedJul 7, 2025

• [bug] extra copy on gpu0 when loading the model on specific GPU

  #15102 closedJul 7, 2025

• PaddleOCR: SyntaxWarning: invalid escape sequence '\W' in ppocr/postprocess/rec_postprocess.py

  #15161 closedJul 7, 2025

• 报错FatalError: `Process abort signal` is detected by the operating system.

  #15475 closedJul 7, 2025

• paddleOCR3.0的BUG

  #15557 closedJul 6, 2025

• paddleocr3.0.0对长条形文字识别存在问题，需要修复！

  #15597 closedJul 6, 2025

9 Issues opened by9 people

• 请问对于字符间隔比较紧密的数字文本识别效果不好 怎么优化呢

  #16027 openedJul 11, 2025

• 识别pdf文件中表格内容

  #16015 openedJul 10, 2025

• ppocrv5 推理显存持续升高

  #16011 openedJul 9, 2025

• 昇腾910B paddle适配报错

  #16010 openedJul 9, 2025

• 识别模型莫名的多出字符“1”

  #16008 openedJul 9, 2025

• PP-Structure v3 推理pdf后 pdf被占用删除失败

  #15999 openedJul 8, 2025

• return_word_box参数出错

  #15981 openedJul 7, 2025

• 算法、图像增强问题

  #15978 openedJul 7, 2025

• 指定输出结果集

  #15973 openedJul 7, 2025

26 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• Export model Error: 'NRTRLabelDecode'

  #15901 commented onJul 6, 2025 • 0 new comments

• mobile rec V5微调后转nb的问题

  #15924 commented onJul 7, 2025 • 0 new comments

• 不支持中文路径和文件名

  #15734 commented onJul 7, 2025 • 0 new comments

• FatalError: `Illegal instruction` is detected by the operating system.

  #15107 commented onJul 7, 2025 • 0 new comments

• 3.0版本训练时无法创建train.log

  #15931 commented onJul 8, 2025 • 0 new comments

• [Regression] Lack of Slice Parameter

  #15868 commented onJul 8, 2025 • 0 new comments

• Paddleocr3.0.0 版本运行报错

  #15558 commented onJul 8, 2025 • 0 new comments

• 公式识别formulaNet-s的onnx模型GPU模型推理非第一次输出异常

  #15125 commented onJul 8, 2025 • 0 new comments

• mac 运行没有结果，没有有效日志。

  #15727 commented onJul 8, 2025 • 0 new comments

• Vote for dependency splitting 依赖切分投票

  #15919 commented onJul 8, 2025 • 0 new comments

• PaddleOCR == 3.0.0 / Enable word-level detection (not line-level)

  #15760 commented onJul 9, 2025 • 0 new comments

• 5060Ti train error

  #15720 commented onJul 9, 2025 • 0 new comments

• 5090显卡是不是不支持？

  #15730 commented onJul 9, 2025 • 0 new comments

• ocr使用gpu识别的时候,不时会有超长耗时出现

  #15166 commented onJul 9, 2025 • 0 new comments

• No Korean Support in PP-OCRv5

  #15373 commented onJul 9, 2025 • 0 new comments

• Paddleocr3.0.0版本运行报错

  #15422 commented onJul 9, 2025 • 0 new comments

• 使用官方/PP-OCRv5_server_det_pretrained模型估计icdar2015数据召回很低

  #15447 commented onJul 9, 2025 • 0 new comments

• Illegal instruction

  #15911 commented onJul 9, 2025 • 0 new comments

• v5 icdar训练精度很低

  #15553 commented onJul 9, 2025 • 0 new comments

• cpp_infer中编译的ppocr不支持现有的PP-OCRv5模型

  #15370 commented onJul 10, 2025 • 0 new comments

• PaddleOCR fails to correctly detect German diacritics (ä, ö, ü, ß) in text recognition

  #14861 commented onJul 10, 2025 • 0 new comments

• ValueError in PP_structureV3

  #15967 commented onJul 11, 2025 • 0 new comments

• paddle3.0.0 c++调用ocr gpu模式下乱码

  #15930 commented onJul 11, 2025 • 0 new comments

• nuitka:ModuleNotFoundError: No module named 'ppocr'

  #15947 commented onJul 11, 2025 • 0 new comments

• 使用paddle-ocr V5的3.0.0版本,微调模型时，卡死在[2025/06/28 17:34:57] ppocr INFO: During the training process, after the 0th iteration, an evaluation is run every 1500 iterations

  #15894 commented onJul 12, 2025 • 0 new comments

• 修复非utf-8平台配置文件解析异常。

  #8336 commented onJul 9, 2025 • 0 new comments