mmdetection 训练踩坑之—— KeyError: ‘segm is not in results‘

错误分析

mmdetection 训练完一个 epoch，计算 coco 指标时报错： KeyError: ‘segm is not in results’，训练退出。
堆栈如下：

06/06 08:59:34 - mmengine - INFO - bbox_mAP_copypaste: 0.029 0.072 0.018 0.001 0.021 0.033
06/06 08:59:34 - mmengine - INFO - Evaluating segm...
Traceback (most recent call last):
  File "./tools/train.py", line 121, in <module>
    main()
  File "./tools/train.py", line 117, in main
    runner.train()
  File "/***/.conda/envs/open-mmlab/lib/python3.8/site-packages/mmengine/runner/runner.py", line 1777, in train
    model = self.train_loop.run()  # type: ignore
  File "/***/.conda/envs/open-mmlab/lib/python3.8/site-packages/mmengine/runner/loops.py", line 102, in run
    self.runner.val_loop.run()
  File "/***/.conda/envs/open-mmlab/lib/python3.8/site-packages/mmengine/runner/loops.py", line 374, in run
    metrics = self.evaluator.evaluate(len(self.dataloader.dataset))
  File "/***/.conda/envs/open-mmlab/lib/python3.8/site-packages/mmengine/evaluator/evaluator.py", line 79, in evaluate
    _results = metric.evaluate(size)
  File "/***/.conda/envs/open-mmlab/lib/python3.8/site-packages/mmengine/evaluator/metric.py", line 133, in evaluate
    _metrics = self.compute_metrics(results)  # type: ignore
  File "/***/mmdetection/mmdet/evaluation/metrics/coco_metric.py", line 450, in compute_metrics
    raise KeyError(f'{metric} is not in results')
KeyError: 'segm is not in results'
WARNING:torch.distributed.elastic.multiprocessing.api:Sending process 3145417 closing signal SIGTERM
WARNING:torch.distributed.elastic.multiprocessing.api:Sending process 3145418 closing signal SIGTERM
WARNING:torch.distributed.elastic.multiprocessing.api:Sending process 3145419 closing signal SIGTERM
WARNING:torch.distributed.elastic.multiprocessing.api:Sending process 3145420 closing signal SIGTERM
WARNING:torch.distributed.elastic.multiprocessing.api:Sending process 3145421 closing signal SIGTERM
WARNING:torch.distributed.elastic.multiprocessing.api:Sending process 3145422 closing signal SIGTERM
WARNING:torch.distributed.elastic.multiprocessing.api:Sending process 3145423 closing signal SIGTERM
ERROR:torch.distributed.elastic.multiprocessing.api:failed (exitcode: 1) local_rank: 0 (pid: 3145416) of binary: /***/.conda/envs/open-mmlab/bin/python
Traceback (most recent call last):
  File "/***/.conda/envs/open-mmlab/lib/python3.8/runpy.py", line 194, in _run_module_as_main
    return _run_code(code, main_globals, None,
  File "/***/.conda/envs/open-mmlab/lib/python3.8/runpy.py", line 87, in _run_code
    exec(code, run_globals)
  File "/***/.conda/envs/open-mmlab/lib/python3.8/site-packages/torch/distributed/launch.py", line 193, in <module>
    main()
  File "/***/.conda/envs/open-mmlab/lib/python3.8/site-packages/torch/distributed/launch.py", line 189, in main
    launch(args)
  File "/***/.conda/envs/open-mmlab/lib/python3.8/site-packages/torch/distributed/launch.py", line 174, in launch
    run(args)
  File "/***/.conda/envs/open-mmlab/lib/python3.8/site-packages/torch/distributed/run.py", line 710, in run
    elastic_launch(
  File "/***/.conda/envs/open-mmlab/lib/python3.8/site-packages/torch/distributed/launcher/api.py", line 131, in __call__

定位代码：mmdetection/mmdet/evaluation/metrics/coco_metric.py:419 行，经调试发现是在计算 segm 指标时报错。

问题解决

因为我的训练任务是 bbox 的目标检测（不是分割任务）所以在配置文件中需要配置 val_evaluator 和 text_evaluator 的 metric 项为：metric=[ ‘bbox’, ], 而默认生成的配置项目 val_evaluator.metric=[ ‘bbox’, ’segm‘, ]。
mmdetection 会在每个轮次训练完成之后，在测试集上进行推理测试，在推理结果上进行相应的指标计算。如：bbox 指标计算是通过在测试集上对推理结果和标注上进行 IoU 计算获得。