[YOLOv8] 问题分析与解决 - RuntimeError: An attempt has been made to start a new process ...bootstrapping ..

问题描述

当使用YOLOv8来训练自定义数据集时，代码如下： 

from ultralytics import YOLO

# 加载一个模型
model = YOLO('yolov8n.yaml')  # 从YAML建立一个新模型
# 训练模型
results = model.train(
    data='D:/YOLOv8Train/v8_train_datasets/mktk_dataset/data.yaml',
    device='0',
    epochs=5,
    batch=4,
    verbose=False,
    imgsz=640)

抛出如下错误：

AMP: running Automatic Mixed Precision (AMP) checks with YOLOv8n...
AMP: checks passed ✅
train: Scanning D:\YOLOv8Train\v8_train_datasets\mktk_dataset\train\labels.cache... 113 images, 0 backgrounds, 0 corrupt: 100%|██████████| 113/113 [00:00<?, ?it/s]
Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "D:\my_project\Anaconda3\envs\yolov8_train\lib\multiprocessing\spawn.py", line 116, in spawn_main
    exitcode = _main(fd, parent_sentinel)
  File "D:\my_project\Anaconda3\envs\yolov8_train\lib\multiprocessing\spawn.py", line 125, in _main
    prepare(preparation_data)
  File "D:\my_project\Anaconda3\envs\yolov8_train\lib\multiprocessing\spawn.py", line 236, in prepare
    _fixup_main_from_path(data['init_main_from_path'])
  File "D:\my_project\Anaconda3\envs\yolov8_train\lib\multiprocessing\spawn.py", line 287, in _fixup_main_from_path
    main_content = runpy.run_path(main_path,
  File "D:\my_project\Anaconda3\envs\yolov8_train\lib\runpy.py", line 289, in run_path
    return _run_module_code(code, init_globals, run_name,
  File "D:\my_project\Anaconda3\envs\yolov8_train\lib\runpy.py", line 96, in _run_module_code
    _run_code(code, mod_globals, init_globals,
  File "D:\my_project\Anaconda3\envs\yolov8_train\lib\runpy.py", line 86, in _run_code
    exec(code, run_globals)
  File "D:\my_project\wepy\src\wepy\aitool\train\yolov8_train.py", line 9, in <module>
    results = model.train(
  File "D:\my_project\Anaconda3\envs\yolov8_train\lib\site-packages\ultralytics\engine\model.py", line 338, in train
    self.trainer.train()
  File "D:\my_project\Anaconda3\envs\yolov8_train\lib\site-packages\ultralytics\engine\trainer.py", line 190, in train
    self._do_train(world_size)
  File "D:\my_project\Anaconda3\envs\yolov8_train\lib\site-packages\ultralytics\engine\trainer.py", line 286, in _do_train
    self._setup_train(world_size)
  File "D:\my_project\Anaconda3\envs\yolov8_train\lib\site-packages\ultralytics\engine\trainer.py", line 251, in _setup_train
    self.train_loader = self.get_dataloader(self.trainset, batch_size=batch_size, rank=RANK, mode='train')
  File "D:\my_project\Anaconda3\envs\yolov8_train\lib\site-packages\ultralytics\models\yolo\detect\train.py", line 52, in get_dataloader
    return build_dataloader(dataset, batch_size, workers, shuffle, rank)  # return dataloader
  File "D:\my_project\Anaconda3\envs\yolov8_train\lib\site-packages\ultralytics\data\build.py", line 107, in build_dataloader
    return InfiniteDataLoader(dataset=dataset,
  File "D:\my_project\Anaconda3\envs\yolov8_train\lib\site-packages\ultralytics\data\build.py", line 33, in __init__
    self.iterator = super().__iter__()
  File "D:\my_project\Anaconda3\envs\yolov8_train\lib\site-packages\torch\utils\data\dataloader.py", line 438, in __iter__
    return self._get_iterator()
  File "D:\my_project\Anaconda3\envs\yolov8_train\lib\site-packages\torch\utils\data\dataloader.py", line 386, in _get_iterator
    return _MultiProcessingDataLoaderIter(self)
  File "D:\my_project\Anaconda3\envs\yolov8_train\lib\site-packages\torch\utils\data\dataloader.py", line 1039, in __init__
    w.start()
  File "D:\my_project\Anaconda3\envs\yolov8_train\lib\multiprocessing\process.py", line 121, in start
    self._popen = self._Popen(self)
  File "D:\my_project\Anaconda3\envs\yolov8_train\lib\multiprocessing\context.py", line 224, in _Popen
    return _default_context.get_context().Process._Popen(process_obj)
  File "D:\my_project\Anaconda3\envs\yolov8_train\lib\multiprocessing\context.py", line 336, in _Popen
    return Popen(process_obj)
  File "D:\my_project\Anaconda3\envs\yolov8_train\lib\multiprocessing\popen_spawn_win32.py", line 45, in __init__
    prep_data = spawn.get_preparation_data(process_obj._name)
  File "D:\my_project\Anaconda3\envs\yolov8_train\lib\multiprocessing\spawn.py", line 154, in get_preparation_data
    _check_not_importing_main()
  File "D:\my_project\Anaconda3\envs\yolov8_train\lib\multiprocessing\spawn.py", line 134, in _check_not_importing_main
    raise RuntimeError('''
RuntimeError: 
        An attempt has been made to start a new process before the
        current process has finished its bootstrapping phase.

        This probably means that you are not using fork to start your
        child processes and you have forgotten to use the proper idiom
        in the main module:

            if __name__ == '__main__':
                freeze_support()
                ...

        The "freeze_support()" line can be omitted if the program
        is not going to be frozen to produce an executable.

原因分析和解决

从错误来看应该是多进程冲突引起的问题，因为训练参数batch设置为4，YOLOv8后台会通过multiprocess模块来启动4个进程进行图像加载处理，对于windows下的python多进程程序，使用的是spawn方式，不是fork方式，我们需要把代码逻辑写到__main__方法中，代码修改如下：

from ultralytics import YOLO

if __name__ == '__main__':
    # 加载一个模型
    model = YOLO('yolov8n.yaml')  # 从YAML建立一个新模型
    # 训练模型
    results = model.train(
        data='D:/YOLOv8Train/v8_train_datasets/mktk_dataset/data.yaml',
        device='0',
        epochs=5,
        batch=4,
        verbose=False,
        imgsz=640)

重新运行代码，问题解决： 

不知道在Linux 会不会有这样的问题，改天有空再试试。