MMDetection3D系列问题解决方法

---

前言

本文档记录了我在学习MMDetection3D过程中遇到的各种BUG和相应的解决方法，亲测有效，希望对大家有所帮助。

环境：

mmcv-full                     1.6.0
mmdet                         2.24.0
mmdet3d                       1.0.0rc4     /home/zzz/mmdetection3d_1.0.0/mmdetection3d-1.0.0rc4
mmsegmentation                0.25.0

MMDetection3D官方教程：MM3d

一、mmcv中出现TypeError: FormatCode() got an unexpected keyword argument ‘verify‘

yapf版本过高，需要降级

pip install yapf==0.40.0

二、AttributeError: module ‘setuptools._distutils’ has no attribute ‘version’

Traceback (most recent call last):
  File "./tools/train.py", line 263, in <module>
    main()
  File "./tools/train.py", line 252, in main
    train_model(
  File "/home/ztl/mmdetection3d_1.0.0/mmdetection3d-1.0.0rc4/mmdet3d/apis/train.py", line 344, in train_model
    train_detector(
  File "/home/ztl/mmdetection3d_1.0.0/mmdetection3d-1.0.0rc4/mmdet3d/apis/train.py", line 319, in train_detector
    runner.run(data_loaders, cfg.workflow)
  File "/home/ztl/anaconda3/envs/mm3d_b/lib/python3.8/site-packages/mmcv/runner/epoch_based_runner.py", line 117, in run
    self.call_hook('before_run')
  File "/home/ztl/anaconda3/envs/mm3d_b/lib/python3.8/site-packages/mmcv/runner/base_runner.py", line 317, in call_hook
    getattr(hook, fn_name)(self)
  File "/home/ztl/anaconda3/envs/mm3d_b/lib/python3.8/site-packages/mmcv/runner/dist_utils.py", line 135, in wrapper
    return func(*args, **kwargs)
  File "/home/ztl/anaconda3/envs/mm3d_b/lib/python3.8/site-packages/mmcv/runner/hooks/logger/tensorboard.py", line 47, in before_run
    from torch.utils.tensorboard import SummaryWriter
  File "/home/ztl/anaconda3/envs/mm3d_b/lib/python3.8/site-packages/torch/utils/tensorboard/__init__.py", line 4, in <module>
    LooseVersion = distutils.version.LooseVersion
AttributeError: module 'setuptools._distutils' has no attribute 'version'

问题：版本过高，‘setuptools._distutils’ 不支持version参数，需要降级。

pip install setuptools==59.5.0

三、 numba 报错 SystemError: initialization of _internal failed without raising an exception

numba 版本和 numpy 版本不兼容

测试得到 numba 0.53.0 和 numpy1.22.1可以解决该问题

pip show 查看 numpy 和 numba 版本

numba                     0.53.0
numpy                     1.22.1

解决方法：重新安装numpy

pip install numpy==1.22.1

注意：numba与mmcv有强依赖，因此不可以改numba版本