compute average training speed and time after training from (.log.json) file.

Thanks for your error report and we appreciate it a lot.

Checklist

1. I have searched related issues but cannot get the expected help. yes
2. I have read the FAQ documentation but cannot get the expected help. yes
3. The bug has not been fixed in the latest version. yes , as far as i know

not display correct value for train time ,see bellow image. Describe the bug A clear and concise description of what the bug is.

i am trying to train CenterNet in my own custom dataset ,which has the same format of COCO dataset. i got the results of the training as (.log.json) and got images tested by a desired (epoch.pth) file . but when i try to analysis the training and find the time of training i got a nan value here is the error massage:

Reproduction

1. What command or script did you run?

python tools/analysis_tools/analyze_logs.py cal_train_time train_results/centernet/20220309_152747.log.json

1. Did you make any modifications on the code or config? Did you understand what you have modified? yes , i made same modification on other models ,for example fasterRCNN , CentriapetalNet , and it works ,and i got the results of training time.
2. What dataset did you use? my custom dataset ,with a same format of COCO one.

Environment

1. Please run python mmdet/utils/collect_env.py to collect necessary environment information and paste it here.

TorchVision: 0.8.0a0 OpenCV: 4.3.0 MMCV: 1.3.8 MMCV Compiler: GCC 7.5 MMCV CUDA Compiler: 11.0 MMDetection: 2.18.1+393c376

1. You may add addition that may be helpful for locating the problem, such as
   • How you installed PyTorch [e.g., pip, conda, source] it was installed before in the server (GPU) using some docker files.
   • Other environment variables that may be related (such as $PATH, $LD_LIBRARY_PATH, $PYTHONPATH, etc.) i install seaborn using this instruction pip install seaborn as the system asked me.

Error traceback If applicable, paste the error trackback here.

tools/analysis_tools/analyze_logs.py:21: RuntimeWarning: Mean of empty slice.
  epoch_ave_time = all_times.mean(-1)
/opt/conda/lib/python3.6/site-packages/numpy/core/_methods.py:163: RuntimeWarning: invalid value encountered in true_divide
  ret, rcount, out=ret, casting='unsafe', subok=False)
/opt/conda/lib/python3.6/site-packages/numpy/core/fromnumeric.py:3373: RuntimeWarning: Mean of empty slice.
  out=out, **kwargs)
/opt/conda/lib/python3.6/site-packages/numpy/core/_methods.py:170: RuntimeWarning: invalid value encountered in double_scalars
  ret = ret.dtype.type(ret / rcount)

A placeholder for trackback.

Bug fix If you have already identified the reason, you can provide the information here. If you are willing to create a PR to fix it, please also leave a comment here and that would be much appreciated!

[jbwang1997]

Yes, I think they are all caused by this reason.

[jbwang1997]

The GPU number is set in the command line. Not displayed in config.