About training and testing on the avenue and ped datasets

[Smallwhite999]

You do a great job and it's a lot of interesting. I found some problems in the process of reproducing your work. After training for 5 epochs on avenue and ped2, an error will be reported after the first verification：

File "main.py", line 128, in train
smoothed_auc, smoothed_auc_avg, temp_timestamp = val(args, net)
File "main.py", line 189, in val
micro_auc, macro_auc = save_and_evaluate(video_output, running_date, dataset=args.dataset)
File "main.py", line 200, in save_and_evaluate
video_output_spatial, video_output_temporal, video_output_complete = remake_video_3d_output(video_output, dataset=dataset)
File "/home/jigsaw/aggregate.py", line 118, in remake_video_3d_output
object_list = load_objects(dataset, frame_num=frame_num)
File "/home/jigsaw/aggregate.py", line 87, in load_objects
detect_result = detect[video_file][frame]
IndexError: list index out of range

but my data structure is the same as yours, but I don’t know what is the reason? Can you answer it, thank you

[zz1565]

Hi，I have the same problem as you. Have you solved it？