Some test videos have already been invalid. How do I deal with these missing videos in result file, fill the value with empty list like “missing_vid”: , or just skip these files?
Also, this line in eval code may cause different val/test set for each submit, because the metric depends on the dictionary keys in predict pkl file.
Thank you for your question. We have updated our codes for evaluation by adding this line: https://github.com/StanLei52/GEBD/blob/619e15c018498adabe951544f803f518a597ac1c/Challenge_eval_Code/eval.py#L25. Thank you for helping us to improve our codes!
In this case you can choose either including or not including the missing videos into your submission file.
Thanks for you quick update!