PaddleOCR/latest/FAQ

[giscus[bot]]

PaddleOCR/latest/FAQ

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/FAQ.html

[qiuhanx]

有没有pgnet使用的详细教程呢？如果有的话可以提供一下网址，十分感谢！

[Zhou2518]

推理报错
[2025/04/12 18:16:18] ppocr INFO: In PP-OCRv3, rec_image_shape parameter defaults to '3, 48, 320', if you are using recognition model with PP-OCRv2 or an older version, please set --rec_image_shape='3,32,320
[2025/04/12 18:16:19] ppocr INFO: Traceback (most recent call last):
File "tools/infer/predict_rec.py", line 860, in main
rec_res, _ = text_recognizer(img_list)
File "tools/infer/predict_rec.py", line 793, in call
self.predictor.run()
OSError: (External) CUDNN error(3000), CUDNN_STATUS_NOT_SUPPORTED.
[Hint: Please search for the error code(3000) on website (https://docs.nvidia.com/deeplearning/cudnn/api/index.html#cudnnStatus_t) to get Nvidia's official solution and advice about CUDNN Error.] (at ../paddle/phi/kernels/fusion/gpu/conv_fusion_kernel.cu:616)
[operator < conv2d_fusion > error]

[2025/04/12 18:16:19] ppocr INFO: (External) CUDNN error(3000), CUDNN_STATUS_NOT_SUPPORTED.
[Hint: Please search for the error code(3000) on website (https://docs.nvidia.com/deeplearning/cudnn/api/index.html#cudnnStatus_t) to get Nvidia's official solution and advice about CUDNN Error.] (at ../paddle/phi/kernels/fusion/gpu/conv_fusion_kernel.cu:616)
[operator < conv2d_fusion > error]

[qiuhanx]

cuda没装好吧？重新安装cuda和cudnn试试 &nbsp; Ash. ***@***.*** &nbsp;

…

------------------&nbsp;原始邮件&nbsp;------------------ 发件人: ***@***.***&gt;; 发送时间: 2025年4月12日(星期六) 晚上6:20 收件人: ***@***.***&gt;; 抄送: ***@***.***&gt;; ***@***.***&gt;; 主题: Re: [PaddlePaddle/PaddleOCR] PaddleOCR/latest/FAQ (Discussion #14722) 推理报错 [2025/04/12 18:16:18] ppocr INFO: In PP-OCRv3, rec_image_shape parameter defaults to '3, 48, 320', if you are using recognition model with PP-OCRv2 or an older version, please set --rec_image_shape='3,32,320 [2025/04/12 18:16:19] ppocr INFO: Traceback (most recent call last): File "tools/infer/predict_rec.py", line 860, in main rec_res, _ = text_recognizer(img_list) File "tools/infer/predict_rec.py", line 793, in call self.predictor.run() OSError: (External) CUDNN error(3000), CUDNN_STATUS_NOT_SUPPORTED. [Hint: Please search for the error code(3000) on website (https://docs.nvidia.com/deeplearning/cudnn/api/index.html#cudnnStatus_t) to get Nvidia's official solution and advice about CUDNN Error.] (at ../paddle/phi/kernels/fusion/gpu/conv_fusion_kernel.cu:616) [operator < conv2d_fusion &gt; error] [2025/04/12 18:16:19] ppocr INFO: (External) CUDNN error(3000), CUDNN_STATUS_NOT_SUPPORTED. [Hint: Please search for the error code(3000) on website (https://docs.nvidia.com/deeplearning/cudnn/api/index.html#cudnnStatus_t) to get Nvidia's official solution and advice about CUDNN Error.] (at ../paddle/phi/kernels/fusion/gpu/conv_fusion_kernel.cu:616) [operator < conv2d_fusion &gt; error] — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you commented.Message ID: ***@***.***&gt;

[jingkuja]

使用华为升腾910b npu，如何使用多卡，或者可以指定设备号启动，目前设置都制定0号设备

[13161360590]

为什么我使用了ocr = PaddleOCR(lang="ch", device="GPU", ocr_version="PP-OCRv5", use_doc_orientation_classify=False, use_doc_unwarping=False, use_textline_orientation=False)，检测出来的还是英文，我的图片内容是中文

[cuicheng01]

ch 指的是支持识别中文的模型，这里用的就是 PP-OCRv5 主模型——PP-OCRv5_server。如果你的图中的确有一些很像英文的中文，确实有极低的概率会识别成英文，可以 show 一下你的 case 看看？