Validate the extracted clips has a small mistake

[fb-reps]

Hello,
As VQ2D/readme.md description:

 bash
# Validate the extracted clips (should take 30 minutes)
python tools/validate_extracted_clips.py \
    --annot-paths data/vq_val.json data/vq_test_unannotated.json \
    --clips-root data/clips

but when I run:

 bash
python tools/validate_extracted_clips.py \
    --annot-paths  data/vq_test_unannotated.json \
    --clips-root data/clips

There will be a error:

 bash
Traceback (most recent call last):
  File "tools/validate_extracted_clips.py", line 95, in 
    main(args)
  File "tools/validate_extracted_clips.py", line 84, in main
    check_clip(c_uid, clips2annotations[c_uid], clip2metadata[c_uid], args)
  File "tools/validate_extracted_clips.py", line 41, in check_clip
    rt_fnos = [rf["frame_number"] for rf in qset["response_track"]]
KeyError: 'response_track'

by the way, when I use path 'data/vq_val.json', it work.

[ga1i13o]

I have the same issue. Any fixes?

[cyfedu-dlut]

Hi, bro, have you solved this issue? i also meet this problem. We can discuss it.