PaddleOCR/latest/version3.x/installation

[giscus[bot]]

PaddleOCR/latest/version3.x/installation

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)

https://paddlepaddle.github.io/PaddleOCR/latest/version3.x/installation.html

[Kisjjw]

python3.8.10版本无法按照，Requires-Python >=3.9

[osborngod]

要不然降paddle版本，要不然升级python版本，自己非要卡bug没意义。

[AuditAIH]

python 3.12

python -m https://paddle-qa.bj.bcebos.com/paddle-pipeline/Develop-TagBuild-Training-Windows-Gpu-Cuda12.9-Cudnn9.9-Trt10.5-Mkl-Avx-VS2019-SelfBuiltPypiUse/86d658f56ebf3a5a7b2b33ace48f22d10680d311/paddlepaddle_gpu-3.0.0.dev20250717-cp312-cp312-win_amd64.whl
python -m pip install 缺少命令
然后我想问一下，有Linux的编译好的支持WHL的包吗，我的操作系统是Windows11 LTSC 下的WSL2 Ubuntu24.04操作系统

[zengqiang331-hash]

5070gpu 的版本，按照指导安装好gpu版本后，gpu利用率达到100%，但是功率只有不到60w。同样的代码只更改device ，cpu运行速度比gpu更快。
请问，是什么问题？

[CocoX-happy]

安装最新paddleocr[all]后，langchain1.0.0执行报错
venv/Lib/site-packages/paddlex/inference/pipelines/components/retriever/base.py
from langchain.docstore.document import Document
ModuleNotFoundError: No module named 'langchain.docstore'
from langchain.text_splitter import RecursiveCharacterTextSplitter
ModuleNotFoundError: No module named 'langchain.text_splitter'

需要修改源文件paddlex/inference/pipelines/components/retriever/base.py
if is_dep_available("langchain"):
from langchain.docstore.document import Document
from langchain.text_splitter import RecursiveCharacterTextSplitter
为
if is_dep_available("langchain"):
from langchain_core.documents import Document
from langchain_text_splitters.character import RecursiveCharacterTextSplitter

[WelcomeToWonderland]

paddlepaddle3与torch冲突

ERROR: pip's dependency resolver does not currently take into account all the packages that are installed. This behaviour is the source of the following dependency conflicts.
paddlepaddle-gpu 3.2.0 requires nvidia-cudnn-cu12==9.5.1.17; platform_system == "Linux" and platform_machine == "x86_64", but you have nvidia-cudnn-cu12 9.10.2.21 which is incompatible.
paddlepaddle-gpu 3.2.0 requires nvidia-cusparselt-cu12==0.6.3; platform_system == "Linux" and platform_machine == "x86_64", but you have nvidia-cusparselt-cu12 0.7.1 which is incompatible.
paddlepaddle-gpu 3.2.0 requires nvidia-nccl-cu12==2.25.1; platform_system == "Linux" and platform_machine == "x86_64", but you have nvidia-nccl-cu12 2.27.5 which is incompatible.

cuda12.6 python 3.10
请问paddlepaddle3与什么版本的torch兼容？或者有什么其他方法可以在同一环境中使用paddlepaddle3和torch？

[yshawen]

你好，我是基于docker安装飞浆和paddleocr，使用PPStructureV3()报错“cannot import name 'fused_rms_norm_ext' from 'paddle.incubate.nn.functional' (/usr/local/lib/python3.10/dist-packages/paddle/incubate/nn/functional/init.py)”，请问是版本不匹配的原因吗，paddlepaddle和paddleocr的版本对应关系应该是什么样的。以下是我电脑环境和安装过程：

[yshawen]

电脑环境：windows，4060，nvidia版本 572.16，dockerb版本 28.5.1
docker安装飞浆：docker run --gpus all --name paddleocr -v D:\pymethods\docker_paddleocr:/paddle --shm-size=8G --network=host -it ccr-2vdh3abv-pub.cnc.bj.baidubce.com/paddlepaddle/paddle:3.0.0-gpu-cuda12.6-cudnn9.5-trt10.5 /bin/bash
安装paddleocr: python -m pip install "paddleocr[all]"
实际安装版本：paddlepaddle-gpu 3.0.0, paddleocr 3.3.1
python version: 3.10.16

[R1310328554]

安装了PaddleOCR 推理包： python -m pip install paddleocr
还需要安装 2.2 安装训练依赖 吗？

[liuhongen1234567]

您好，如果不需要训练，可以只安装推理依赖。

[SilentShadowZWL]

请问pyinstaller打包有没有paddleocr开发打包的说明，总是报错缺失

[yzy726]

在虚拟环境中安装时出现：WARNING: Retrying (Retry(total=4, connect=None, read=None, redirect=None, status=None)) after connection broken by 'SSLError(SSLCertVerificationError('根据当前系统时钟或签名文件中
的时间戳验证时要求的证书不在有效期内。'))': /packages/stable/cu126/paddlepaddle-gpu/