PaddleX/latest/module_usage/tutorials/ocr_modules/text_detection

[giscus[bot]]

PaddleX/latest/module_usage/tutorials/ocr_modules/text_detection

https://paddlepaddle.github.io/PaddleX/latest/module_usage/tutorials/ocr_modules/text_detection.html

[JiaZeShi]

使用API调用时，返回参数里的 ocrImage 会影响识别时间吗，如果影响怎么去掉

[Bobholamovic]

结果可视化过程本身以及在服务端与客户端之间传递可视化结果可能带来额外的开销，通常这部分开销在总响应时间里的占比不高，不过，如果你需要优化掉这部分开销，目前我们提供一个实验性的接口（尚未在文档中体现）：

在产线配置文件中，设置如下项：

Serving:
    visualize: False

这样，结果中将不包含ocrImage等可视化结果字段。

[CrazyDJDave]

paddlex.utils.errors.others.UnsupportedParamError: 'PP-OCRv4_mobile_det' is not a registered model name.
这是什么问题啊

[cuicheng01]

您好，是什么命令报的错呢？

[SophieRoyal]

我的paddlepaddle=3.0.0，paddlex=3.0.3，paddleocr=3.0.3，python环境为3.10。
运行数据效验(
python main.py -c paddlex/configs/modules/text_detection/PP-OCRv5_mobile_det.yaml
-o Global.mode=check_dataset
-o Global.dataset_dir=./dataset/ocr_det_dataset_examples
)，报以下错误：
【Traceback (most recent call last):
File "D:\PycharmProjects\PaddleX\paddlex\utils\result_saver.py", line 28, in wrap
result = func(self, *args, **kwargs)
File "D:\PycharmProjects\PaddleX\paddlex\engine.py", line 38, in run
return self.model.check_dataset()
File "D:\PycharmProjects\PaddleX\paddlex\model.py", line 115, in check_dataset
return dataset_checker.check()
File "D:\PycharmProjects\PaddleX\paddlex\modules\base\dataset_checker\dataset_checker.py", line 81, in check
analysis = self.analyse(dataset_dir)
File "D:\PycharmProjects\PaddleX\paddlex\modules\text_detection\dataset_checker_init.py", line 91, in analyse
return deep_analyse(dataset_dir, self.output)
File "D:\PycharmProjects\PaddleX\paddlex\utils\deps.py", line 135, in _wrapper
return func(*args, **kwargs)
File "D:\PycharmProjects\PaddleX\paddlex\modules\text_detection\dataset_checker\dataset_src\analyse_dataset.py", line 193, in deep_analyse
bar_array = np.frombuffer(canvas.tostring_rgb(), dtype="uint8").reshape(
AttributeError: 'FigureCanvasAgg' object has no attribute 'tostring_rgb'
】

[SophieRoyal]

已找到源码修好了。是源码有这个问题。

替换 bar_array 的生成方式

canvas.draw()
width, height = fig.get_size_inches() * fig.get_dpi()
rgba_array = np.asarray(canvas.buffer_rgba(), dtype="uint8").reshape(int(height), int(width), 4)
bar_array = cv2.cvtColor(rgba_array, cv2.COLOR_RGBA2BGR)

替换 pie_array 的生成方式

canvas.draw()
width, height = fig.get_size_inches() * fig.get_dpi()
rgba_array = np.asarray(canvas.buffer_rgba(), dtype="uint8").reshape(int(height), int(width), 4)
pie_array = cv2.cvtColor(rgba_array, cv2.COLOR_RGBA2BGR)

[congyao123456]

使用官方给的数据集进行文本检测训练demo时候报错：ValueError: I/O operation on closed pipe
025/08/14 00:07:11] ppocr INFO: train with paddle 3.0.0 and device Place(gpu:0)
[2025/08/14 00:07:11] ppocr INFO: Initialize indexes of datasets:['C:\Users\sunjian\PycharmProjects\PaddleX-release-3.1\ocr_det_dataset_examples\ocr_det_dataset_examples\train.txt']
[2025/08/14 00:07:11] ppocr INFO: Initialize indexes of datasets:['C:\Users\sunjian\PycharmProjects\PaddleX-release-3.1\ocr_det_dataset_examples\ocr_det_dataset_examples\val.txt']
W0814 00:07:12.024827 21240 gpu_resources.cc:119] Please NOTE: device: 0, GPU Compute Capability: 8.9, Driver API Version: 12.6, Runtime API Version: 12.6
W0814 00:07:12.025868 21240 gpu_resources.cc:164] device: 0, cuDNN Version: 9.5.
INFO:root:If regularizer of a Parameter has been set by 'paddle.ParamAttr' or 'static.WeightNormParamAttr' already. The weight_decay[1e-06] in Optimizer will not take effect, and it will only be applied to other Parameters!
[2025/08/14 00:07:12] ppocr INFO: train dataloader has 50 iters
[2025/08/14 00:07:12] ppocr INFO: valid dataloader has 50 iters
download https://paddle-model-ecology.bj.bcebos.com/paddlex/official_pretrained_model/PP-OCRv5_mobile_det_pretrained.pdparams to C:\Users\sunjian/.paddleocr/models\PP-OCRv5_mobile_det_pretrained.pdparams
[2025/08/14 00:07:12] ppocr INFO: Path C:\Users\sunjian/.paddleocr/models\PP-OCRv5_mobile_det_pretrained.pdparams already exists. Skipping...
[2025/08/14 00:07:12] ppocr INFO: load pretrain successful from C:\Users\sunjian/.paddleocr/models\PP-OCRv5_mobile_det_pretrained
[2025/08/14 00:07:12] ppocr INFO: During the training process, after the 0th iteration, an evaluation is run every 50 iterations
[2025/08/14 00:07:15] ppocr INFO: epoch: [1/100], global_step: 10, lr: 0.000045, loss: 2.681390, loss_shrink_maps: 1.766998, loss_threshold_maps: 0.558334, loss_binary_maps: 0.353621, loss_cbn: 0.000000, avg_reader_cost: 0.01895 s, avg_batch_cost: 0.31436 s, avg_samples: 4.0, ips: 12.72409 samples/s, eta: 0:26:08, max_mem_reserved: 2924 MB, max_mem_allocated: 2689 MB
[2025/08/14 00:07:17] ppocr INFO: epoch: [1/100], global_step: 20, lr: 0.000095, loss: 2.130130, loss_shrink_maps: 1.270447, loss_threshold_maps: 0.523089, loss_binary_maps: 0.254630, loss_cbn: 0.000000, avg_reader_cost: 0.00208 s, avg_batch_cost: 0.22012 s, avg_samples: 4.0, ips: 18.17150 samples/s, eta: 0:22:10, max_mem_reserved: 2924 MB, max_mem_allocated: 2689 MB
[2025/08/14 00:07:19] ppocr INFO: epoch: [1/100], global_step: 30, lr: 0.000195, loss: 1.575355, loss_shrink_maps: 0.910293, loss_threshold_maps: 0.550951, loss_binary_maps: 0.182273, loss_cbn: 0.000000, avg_reader_cost: 0.00195 s, avg_batch_cost: 0.21857 s, avg_samples: 4.0, ips: 18.30078 samples/s, eta: 0:20:47, max_mem_reserved: 2924 MB, max_mem_allocated: 2689 MB
[2025/08/14 00:07:22] ppocr INFO: epoch: [1/100], global_step: 40, lr: 0.000295, loss: 1.500061, loss_shrink_maps: 0.826224, loss_threshold_maps: 0.506551, loss_binary_maps: 0.164522, loss_cbn: 0.000000, avg_reader_cost: 0.00306 s, avg_batch_cost: 0.22368 s, avg_samples: 4.0, ips: 17.88304 samples/s, eta: 0:20:11, max_mem_reserved: 2924 MB, max_mem_allocated: 2689 MB
[2025/08/14 00:07:24] ppocr INFO: Export inference config file to C:\Users\sunjian\PycharmProjects\PaddleX-release-3.1\output\latest\inference\inference.yml
Skipping import of the encryption module
[2025/08/14 00:07:27] ppocr INFO: inference model is saved to C:\Users\sunjian\PycharmProjects\PaddleX-release-3.1\output\latest\inference\inference
[2025/08/14 00:07:27] ppocr INFO: Already save model info in C:\Users\sunjian\PycharmProjects\PaddleX-release-3.1\output\latest
[2025/08/14 00:07:27] ppocr INFO: save model in C:\Users\sunjian\PycharmProjects\PaddleX-release-3.1\output\latest\latest
[2025/08/14 00:07:27] ppocr INFO: Export inference config file to C:\Users\sunjian\PycharmProjects\PaddleX-release-3.1\output\iter_epoch_1\inference\inference.yml
Skipping import of the encryption module
[2025/08/14 00:07:31] ppocr INFO: inference model is saved to C:\Users\sunjian\PycharmProjects\PaddleX-release-3.1\output\iter_epoch_1\inference\inference
[2025/08/14 00:07:31] ppocr INFO: Already save model info in C:\Users\sunjian\PycharmProjects\PaddleX-release-3.1\output\iter_epoch_1
[2025/08/14 00:07:31] ppocr INFO: save model in C:\Users\sunjian\PycharmProjects\PaddleX-release-3.1\output\iter_epoch_1\iter_epoch_1
[2025/08/14 00:07:32] ppocr INFO: epoch: [2/100], global_step: 50, lr: 0.000395, loss: 1.543667, loss_shrink_maps: 0.887902, loss_threshold_maps: 0.544873, loss_binary_maps: 0.175785, loss_cbn: 0.000000, avg_reader_cost: 0.76120 s, avg_batch_cost: 1.00970 s, avg_samples: 4.0, ips: 3.96158 samples/s, eta: 0:32:46, max_mem_reserved: 2924 MB, max_mem_allocated: 2690 MB
Traceback (most recent call last):
File "C:\Users\sunjian\PycharmProjects\PaddleX-release-3.1\paddlex\utils\result_saver.py", line 28, in wrap
result = func(self, *args, **kwargs)
File "C:\Users\sunjian\PycharmProjects\PaddleX-release-3.1\paddlex\engine.py", line 41, in run
self._model.train()
File "C:\Users\sunjian\PycharmProjects\PaddleX-release-3.1\paddlex\model.py", line 119, in train
trainer.train()
File "C:\Users\sunjian\PycharmProjects\PaddleX-release-3.1\paddlex\modules\base\trainer.py", line 99, in train
train_result = self.pdx_model.train(**train_args)
File "C:\Users\sunjian\PycharmProjects\PaddleX-release-3.1\paddlex\repo_apis\PaddleOCR_api\text_rec\model.py", line 147, in train
return self.runner.train(
File "C:\Users\sunjian\PycharmProjects\PaddleX-release-3.1\paddlex\repo_apis\PaddleOCR_api\text_rec\runner.py", line 55, in train
return self.run_cmd(
File "C:\Users\sunjian\PycharmProjects\PaddleX-release-3.1\paddlex\repo_apis\base\runner.py", line 345, in run_cmd
retcode = asyncio.run(
File "C:\Users\sunjian\Anaconda3\envs\paddle_kf\lib\asyncio\runners.py", line 44, in run
return loop.run_until_complete(main)
File "C:\Users\sunjian\Anaconda3\envs\paddle_kf\lib\asyncio\base_events.py", line 616, in run_until_complete
return future.result()
File "C:\Users\sunjian\PycharmProjects\PaddleX-release-3.1\paddlex\repo_apis\base\runner.py", line 310, in _tee_proc_call
await asyncio.gather(
File "C:\Users\sunjian\PycharmProjects\PaddleX-release-3.1\paddlex\repo_apis\base\runner.py", line 298, in _read_display_and_record_from_stream
line = line.decode(_ENCODING)
UnicodeDecodeError: 'gbk' codec can't decode byte 0x8f in position 20: illegal multibyte sequence
Exception ignored in: <function BaseSubprocessTransport.del at 0x00000208400D4670>
Traceback (most recent call last):
File "C:\Users\sunjian\Anaconda3\envs\paddle_kf\lib\asyncio\base_subprocess.py", line 126, in del
self.close()
File "C:\Users\sunjian\Anaconda3\envs\paddle_kf\lib\asyncio\base_subprocess.py", line 104, in close
proto.pipe.close()
File "C:\Users\sunjian\Anaconda3\envs\paddle_kf\lib\asyncio\proactor_events.py", line 108, in close
self._loop.call_soon(self._call_connection_lost, None)
File "C:\Users\sunjian\Anaconda3\envs\paddle_kf\lib\asyncio\base_events.py", line 719, in call_soon
self._check_closed()
File "C:\Users\sunjian\Anaconda3\envs\paddle_kf\lib\asyncio\base_events.py", line 508, in _check_closed
raise RuntimeError('Event loop is closed')
RuntimeError: Event loop is closed
Exception ignored in: <function _ProactorBasePipeTransport.del at 0x00000208400F2CA0>
Traceback (most recent call last):
File "C:\Users\sunjian\Anaconda3\envs\paddle_kf\lib\asyncio\proactor_events.py", line 115, in del
_warn(f"unclosed transport {self!r}", ResourceWarning, source=self)
File "C:\Users\sunjian\Anaconda3\envs\paddle_kf\lib\asyncio\proactor_events.py", line 79, in repr
info.append(f'fd={self._sock.fileno()}')
File "C:\Users\sunjian\Anaconda3\envs\paddle_kf\lib\asyncio\windows_utils.py", line 102, in fileno
raise ValueError("I/O operation on closed pipe")
ValueError: I/O operation on closed pipe
Exception ignored in: <function _ProactorBasePipeTransport.del at 0x00000208400F2CA0>
Traceback (most recent call last):
File "C:\Users\sunjian\Anaconda3\envs\paddle_kf\lib\asyncio\proactor_events.py", line 115, in del
_warn(f"unclosed transport {self!r}", ResourceWarning, source=self)
File "C:\Users\sunjian\Anaconda3\envs\paddle_kf\lib\asyncio\proactor_events.py", line 79, in repr
info.append(f'fd={self._sock.fileno()}')
File "C:\Users\sunjian\Anaconda3\envs\paddle_kf\lib\asyncio\windows_utils.py", line 102, in fileno
raise ValueError("I/O operation on closed pipe")
ValueError: I/O operation on closed pipe

[wyd1520]

我执行这个命令训练为什么会提示这个错误
python main.py -c paddlex/configs/modules/text_detection/PP-OCRv5_mobile_det.yaml -o Global.mode=train -o Global.dataset_dir=./dataset/ocr_det_dataset_examples

Traceback (most recent call last):
File "G:\PythonDemo\python_paddlex\venv\lib\site-packages\paddlex\repo_apis\base\config.py", line 30, in _create_config
model_info = get_registered_model_info(model_name)
File "G:\PythonDemo\python_paddlex\venv\lib\site-packages\paddlex\repo_apis\base\register.py", line 78, in query
return self._table[prim_key]
KeyError: 'PP-OCRv5_mobile_det'

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
File "G:\PythonDemo\python_paddlex\venv\lib\site-packages\paddlex\utils\result_saver.py", line 28, in wrap
result = func(self, *args, **kwargs)
File "G:\PythonDemo\python_paddlex\venv\lib\site-packages\paddlex\engine.py", line 41, in run
self._model.train()
File "G:\PythonDemo\python_paddlex\venv\lib\site-packages\paddlex\model.py", line 118, in train
trainer = build_trainer(self._config)
File "G:\PythonDemo\python_paddlex\venv\lib\site-packages\paddlex\modules\base\trainer.py", line 44, in build_trainer
return BaseTrainer.get(model_name)(config)
File "G:\PythonDemo\python_paddlex\venv\lib\site-packages\paddlex\modules\base\trainer.py", line 66, in init
self.pdx_config, self.pdx_model = build_model(
File "G:\PythonDemo\python_paddlex\venv\lib\site-packages\paddlex\modules\base\build_model.py", line 31, in build_model
config = Config(model_name, config_path)
File "G:\PythonDemo\python_paddlex\venv\lib\site-packages\paddlex\repo_apis\base\config.py", line 32, in _create_config
raise UnsupportedParamError(
paddlex.utils.errors.others.UnsupportedParamError: 'PP-OCRv5_mobile_det' is not a registered model name.