latest/version3.x/pipeline_usage/PaddleOCR-VL-NPU

[giscus[bot]]

latest/version3.x/pipeline_usage/PaddleOCR-VL-NPU

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://www.paddleocr.ai/latest/version3.x/pipeline_usage/PaddleOCR-VL-NPU.html

[YXlukarov]

镜像没有适配arm架构吗？运行docker命令的时候出错了

docker run -it \
  --user root \
  --name lj_paddleocr_npu \
  --privileged \
  --platform linux/amd64 \  # 新增：指定镜像架构
  -v /usr/local/Ascend/driver:/usr/local/Ascend/driver \
  -v /usr/local/bin/npu-smi:/usr/local/bin/npu-smi \
  -v /usr/local/dcmi:/usr/local/dcmi \
  -v /data1/lj:/data1/lj \
  --shm-size 64g \
  --network host \
  ccr-2vdh3abv-pub.cnc.bj.baidubce.com/paddlepaddle/paddleocr-genai-vllm-server:latest-npu \
  /bin/bash

错误为：The requested inage's platforn (Linu/and64) does not natch the detected host platforn (linu/arnf4/v8) and no specific platforn was requested

[Bobholamovic]

是有的，可以参考最新的文档（tag suffix有变化）

[wangxiang0722]

ARM CPU，加载PP-LCNet_x1_0_doc_ori模型报错，报错内容：

--------------------------------------
C++ Traceback (most recent call last):
--------------------------------------
0   paddle_infer::Predictor::Predictor(paddle::AnalysisConfig const&)
1   std::unique_ptr<paddle::PaddlePredictor, std::default_delete<paddle::PaddlePredictor> > paddle::CreatePaddlePredictor<paddle::AnalysisConfig, (paddle::PaddleEngineKind)2>(paddle::AnalysisConfig const&)
2   paddle::AnalysisPredictor::Init(std::shared_ptr<paddle::framework::Scope> const&, std::shared_ptr<paddle::framework::ProgramDesc> const&)
3   paddle::AnalysisPredictor::PreparePirProgram()
4   paddle::AnalysisPredictor::SaveOrLoadPirParameters(bool)
5   std::filesystem::__cxx11::path::~path()
6   std::vector<std::filesystem::__cxx11::path::_Cmpt, std::allocator<std::filesystem::__cxx11::path::_Cmpt> >::~vector()
7   std::vector<std::filesystem::__cxx11::path::_Cmpt, std::allocator<std::filesystem::__cxx11::path::_Cmpt> >::~vector()

----------------------
Error Message Summary:
----------------------
FatalError: `Segmentation fault` is detected by the operating system.
  [TimeInfo: *** Aborted at 1769685800 (unix time) try "date -d @1769685800" if you are using GNU date ***]
  [SignalInfo: *** SIGSEGV (@0x6623) received by PID 26147 (TID 0xffff81fc2020) from PID 26147 ***]

Segmentation fault (core dumped)

Python代码：

pipeline = PaddleOCRVL(vl_rec_backend="vllm-server", vl_rec_server_url=my_url,
    layout_detection_model_dir=layout_detection_model_dir,
    doc_orientation_classify_model_name="PP-LCNet_x1_0_doc_ori",
    use_doc_orientation_classify=True,
    doc_orientation_classify_model_dir=doc_orientation_classify_model_dir)

已安装的包版本

paddle-custom-npu      3.2.0
paddleocr              3.3.1
paddlepaddle           3.3.0
paddlex                3.3.13

[Bobholamovic]

可以试试我们的NPU官方镜像

[imkui]

Docker没有CPU的镜像吗

[Bobholamovic]

我们强烈不推荐使用CPU部署，因此也没有提供CPU版本的镜像。对于 x64 CPU，手动安装也十分方便，可以尝试～

[WinddyAkoky]

v版面检测排序模型的微调什么时候能支持呀？