连续推理文件会报错

[bcsy-ops]

2025-07-14 15:56:45,654 - main - INFO - 开始处理文件: Temp\fefc2c16-05eb-460d-beaa-10fc305b5a3e\zzz.jpg
2025-07-14 15:56:47,054 - main - ERROR - OCR处理失败: Unknown exception
Traceback (most recent call last):
File "OCRcopy.py", line 73, in ocr1
result = ocr.predict(input=input_file)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "anaconda3\envs\OCR\Lib\site-packages\paddleocr_pipelines\ocr.py", line 208, in predict
return list(
^^^^^
File "anaconda3\envs\OCR\Lib\site-packages\paddlex\inference\pipelines_parallel.py", line 129, in predict
yield from self._pipeline.predict(
File "anaconda3\envs\OCR\Lib\site-packages\paddlex\inference\pipelines\ocr\pipeline.py", line 350, in predict
det_results = list(
^^^^^
File "anaconda3\envs\OCR\Lib\site-packages\paddlex\inference\models\base\predictor\base_predictor.py", line 211, in call
yield from self.apply(input, **kwargs)
File "anaconda3\envs\OCR\Lib\site-packages\paddlex\inference\models\base\predictor\base_predictor.py", line 267, in apply
prediction = self.process(batch_data, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "anaconda3\envs\OCR\Lib\site-packages\paddlex\inference\models\text_detection\predictor.py", line 105, in process
batch_preds = self.infer(x=x)
^^^^^^^^^^^^^^^
File "anaconda3\envs\OCR\Lib\site-packages\paddlex\inference\models\common\static_infer.py", line 287, in call
pred = self.infer(x)
^^^^^^^^^^^^^
File "anaconda3\envs\OCR\Lib\site-packages\paddlex\inference\models\common\static_infer.py", line 252, in call
self.predictor.run()
RuntimeError: Unknown exception

使用的Flask封装成一个接口，并在全局变量中加载本地模型实例，调用接口后第一次推理可出结果，但后续使用就会报该错误，在不关闭该接口的一段时间后可重新推理一次。
PaddleOCR 版本: 3.1.0
PaddlePaddle 版本: 3.1.0
python 3.12.11

[mysc6699]

我这边也遇到一样的问题。后面测试每个线程创建自己的 pipeline对象后有所改善

[Bobholamovic]

请参考 #15621