PP-OCR didn't work on GPU...

[mkrzywda]

🔎 Search before asking

• I have searched the PaddleOCR Docs and found no similar bug report.
• I have searched the PaddleOCR Issues and found no similar bug report.
• I have searched the PaddleOCR Discussions and found no similar bug report.

🐛 Bug (问题描述)

I have a issue when I try run PP-OCR instead of default version PP-OCRv4

==========
== CUDA ==
==========

CUDA Version 12.2.0

Container image Copyright (c) 2016-2023, NVIDIA CORPORATION & AFFILIATES. All rights reserved.

This container image and its contents are governed by the NVIDIA Deep Learning Container License.
By pulling and using the container, you accept the terms and conditions of this license:
https://developer.nvidia.com/ngc/nvidia-deep-learning-container-license

A copy of this license is made available in this container at /NGC-DL-CONTAINER-LICENSE for your convenience.

[2025/02/18 12:40:08] ppocr DEBUG: Namespace(help='==SUPPRESS==', use_gpu=True, use_xpu=False, use_npu=False, use_mlu=False, ir_optim=True, use_tensorrt=False, min_subgraph_size=15, precision='fp32', gpu_mem=500, gpu_id=0, image_dir=None, page_num=0, det_algorithm='DB', det_model_dir='/root/.paddleocr/whl/det/en/en_ppocr_mobile_v2.0_det_infer', det_limit_side_len=960, det_limit_type='max', det_box_type='quad', det_db_thresh=0.3, det_db_box_thresh=0.6, det_db_unclip_ratio=1.5, max_batch_size=10, use_dilation=False, det_db_score_mode='fast', det_east_score_thresh=0.8, det_east_cover_thresh=0.1, det_east_nms_thresh=0.2, det_sast_score_thresh=0.5, det_sast_nms_thresh=0.2, det_pse_thresh=0, det_pse_box_thresh=0.85, det_pse_min_area=16, det_pse_scale=1, scales=[8, 16, 32], alpha=1.0, beta=1.0, fourier_degree=5, rec_algorithm='SVTR_LCNet', rec_model_dir='/root/.paddleocr/whl/rec/en/en_number_mobile_v2.0_rec_infer', rec_image_inverse=True, rec_image_shape='3, 32, 320', rec_batch_num=6, max_text_length=100, rec_char_dict_path='/usr/local/lib/python3.10/dist-packages/paddleocr/ppocr/utils/en_dict.txt', use_space_char=True, vis_font_path='./doc/fonts/simfang.ttf', drop_score=0.5, e2e_algorithm='PGNet', e2e_model_dir=None, e2e_limit_side_len=768, e2e_limit_type='max', e2e_pgnet_score_thresh=0.5, e2e_char_dict_path='./ppocr/utils/ic15_dict.txt', e2e_pgnet_valid_set='totaltext', e2e_pgnet_mode='fast', use_angle_cls=False, cls_model_dir='/root/.paddleocr/whl/cls/ch_ppocr_mobile_v2.0_cls_infer', cls_image_shape='3, 48, 192', label_list=['0', '180'], cls_batch_num=6, cls_thresh=0.9, enable_mkldnn=False, cpu_threads=10, use_pdserving=False, warmup=False, sr_model_dir=None, sr_image_shape='3, 32, 128', sr_batch_num=1, draw_img_save_dir='./inference_results', save_crop_res=False, crop_res_save_dir='./output', use_mp=False, total_process_num=1, process_id=0, benchmark=False, save_log_path='./log_output/', show_log=True, use_onnx=False, return_word_box=False, output='./output', table_max_len=488, table_algorithm='TableAttn', table_model_dir=None, merge_no_span_structure=True, table_char_dict_path=None, formula_algorithm='LaTeXOCR', formula_model_dir=None, formula_char_dict_path=None, formula_batch_num=1, layout_model_dir=None, layout_dict_path=None, layout_score_threshold=0.5, layout_nms_threshold=0.5, kie_algorithm='LayoutXLM', ser_model_dir=None, re_model_dir=None, use_visual_backbone=True, ser_dict_path='../train_data/XFUND/class_list_xfun.txt', ocr_order_method=None, mode='structure', image_orientation=False, layout=True, table=True, formula=False, ocr=True, recovery=False, recovery_to_markdown=False, use_pdf2docx_api=False, invert=False, binarize=False, alphacolor=(255, 255, 255), lang='en', det=True, rec=True, type='ocr', savefile=False, **ocr_version='PP-OCR'**, structure_version='PP-StructureV2', cls=False, bin=False, inv=False, alpha_color=False)
[2025/02/18 12:40:08] ppocr WARNING: The first GPU is used for inference by default, GPU ID: 0
[2025/02/18 12:40:08] ppocr WARNING: The first GPU is used for inference by default, GPU ID: 0
2025-02-18 12:40:08,783 - INFO - Server started, listening on port 8000
2025-02-18 12:40:08,783 - INFO - Prometheus metrics available on port 8001
[2025/02/18 12:40:17] ppocr WARNING: Since the angle classifier is not initialized, it will not be used during the forward process
[2025/02/18 12:40:17] ppocr DEBUG: dt_boxes num : 60, elapsed : 0.2818129062652588


--------------------------------------
C++ Traceback (most recent call last):
--------------------------------------
No stack trace in paddle, may be caused by external reasons.

----------------------
Error Message Summary:
----------------------
FatalError: Segmentation fault is detected by the operating system.
  [TimeInfo: *** Aborted at 1739882417 (unix time) try "date -d @1739882417" if you are using GNU date ***]
  [SignalInfo: *** SIGSEGV (@0x0) received by PID 28 (TID 0x784cc7e00640) from PID 0 ***]

/app/startup.sh: line 4:    28 Segmentation fault      (core dumped) 

🏃‍♂️ Environment (运行环境)

Dockerfile:

FROM nvidia/cuda:12.2.0-devel-ubuntu22.04

RUN apt-get update -y \
    && apt-get install -y python3-pip \
    && apt-get install -y less \
    && apt-get install -y libgl1 \
    && apt-get install -y libglib2.0-0 \
    && apt-get install -y cudnn9-cuda-12

WORKDIR app

COPY requirements.txt requirements.txt
RUN pip install -r requirements.txt


RUN python3 -c "from paddleocr import PaddleOCR; PaddleOCR(use_angle_cls=False, lang='en',  ocr_version='PP-OCR', use_space_char=True, cls=False, bin=False, inv=False, alpha_color=False, max_text_length=100)"

COPY startup.sh startup.sh

RUN chmod +x startup.sh

EXPOSE 8000
ENTRYPOINT ["/app/startup.sh"]

startup.sh

#!/bin/bash
/opt/nvidia/nvidia_entrypoint.sh
python3 server.py

But when I build for CPU:

FROM python:3.12-slim

WORKDIR app

RUN apt-get update && apt-get install -y --no-install-recommends libgomp1 \
        libgl1 \
        libglib2.0-0

COPY requirements-cpu.txt requirements.txt
RUN pip install -r requirements.txt

RUN python3 -c "from paddleocr import PaddleOCR; PaddleOCR(use_angle_cls=False, lang='en',  ocr_version='PP-OCR', use_space_char=True, cls=False, bin=False, inv=False, alpha_color=False, max_text_length=100)"

COPY startup.sh startup.sh

RUN chmod +x startup.sh

EXPOSE 8000
ENTRYPOINT ["/app/startup.sh"]

works correctly.

🌰 Minimal Reproducible Example (最小可复现问题的Demo)

ocr = PaddleOCR(
    use_angle_cls=False,
    lang="en",
    use_gpu=True,
    ocr_version="PP-OCR",
    use_space_char=True,
    cls=False,
    bin=False,
    inv=False,
    alpha_color=False,
    max_text_length=100,
)

requirements for GPU:

pyclipper==1.3.0.post6
attrdict==2.0.1
albucore==0.0.20
albumentations==1.4.21
paddlepaddle-gpu==2.6.2
paddleocr==2.9.0
grpcio==1.68.1
grpcio-tools==1.68.1
prometheus_client==0.21.0
ruff==0.9.5

requirements.txt for CPU:

pyclipper==1.3.0.post5
attrdict==2.0.1
albucore==0.0.16
albumentations==1.4.14
paddlepaddle==3.0.0b0
paddleocr==2.9.0
grpcio==1.68.0
grpcio-tools==1.68.0
prometheus_client==0.21.0
ruff==0.9.5

[GreatV]

从你提供的错误信息来看，PP-OCR在GPU上运行时出现了段错误（Segmentation fault）。这个问题可能与以下几个原因有关：

1. CUDA版本不兼容：你使用的CUDA版本是12.2.0，而PaddlePaddle的某些版本可能对CUDA版本有特定的要求。建议检查PaddlePaddle的官方文档，确认你使用的paddlepaddle-gpu==2.6.2是否支持CUDA 12.2.0。如果不支持，建议降级CUDA版本或升级PaddlePaddle到支持CUDA 12.2.0的版本。

2. GPU驱动问题：确保你的GPU驱动是最新的，并且与CUDA版本兼容。你可以通过nvidia-smi命令检查GPU驱动和CUDA版本是否匹配。

3. PaddlePaddle安装问题：在Dockerfile中，你安装了paddlepaddle-gpu==2.6.2，但有时安装过程中可能会出现依赖问题。建议在安装PaddlePaddle之前，先安装nvidia-cudnn和nvidia-cuda-toolkit，以确保所有必要的库都已正确安装。

4. Docker环境配置：确保Docker容器能够正确访问GPU。你可以通过nvidia-docker来运行容器，以确保GPU资源被正确分配。

5. PaddleOCR版本问题：你使用的PaddleOCR版本是2.9.0，建议检查是否有已知的与GPU相关的bug。如果有，建议升级到最新版本或回退到一个稳定的版本。

解决方案建议

1. 检查CUDA和PaddlePaddle的兼容性：

   • 确认paddlepaddle-gpu==2.6.2是否支持CUDA 12.2.0。如果不支持，建议降级CUDA版本或升级PaddlePaddle。

2. 更新GPU驱动：

   • 确保GPU驱动是最新的，并且与CUDA版本兼容。

3. 重新安装PaddlePaddle：

   • 在Dockerfile中，确保安装了所有必要的依赖库，如nvidia-cudnn和nvidia-cuda-toolkit。

4. 使用nvidia-docker：

   • 使用nvidia-docker来运行容器，以确保GPU资源被正确分配。

5. 检查PaddleOCR版本：

   • 检查是否有已知的与GPU相关的bug，并考虑升级或回退PaddleOCR版本。

示例Dockerfile修改

FROM nvidia/cuda:12.2.0-devel-ubuntu22.04

RUN apt-get update -y \
    && apt-get install -y python3-pip \
    && apt-get install -y less \
    && apt-get install -y libgl1 \
    && apt-get install -y libglib2.0-0 \
    && apt-get install -y cudnn9-cuda-12

WORKDIR app

COPY requirements.txt requirements.txt
RUN pip install -r requirements.txt

# 确保安装了nvidia-cudnn和nvidia-cuda-toolkit
RUN apt-get install -y nvidia-cudnn nvidia-cuda-toolkit

RUN python3 -c "from paddleocr import PaddleOCR; PaddleOCR(use_angle_cls=False, lang='en',  ocr_version='PP-OCR', use_space_char=True, cls=False, bin=False, inv=False, alpha_color=False, max_text_length=100)"

COPY startup.sh startup.sh

RUN chmod +x startup.sh

EXPOSE 8000
ENTRYPOINT ["/app/startup.sh"]

运行Docker容器

nvidia-docker run -it --rm your-docker-image

通过以上步骤，你应该能够解决PP-OCR在GPU上运行时出现的段错误问题。如果问题仍然存在，建议进一步检查系统日志或联系PaddlePaddle社区获取支持。

---

Response generated by 🤖 feifei-bot | deepseek-chat

⚠️ This is an AI-generated response. Please verify any critical information.