ASDNet

ASDNet copied to clipboard

After running data/extract_face_crops_time.py, there were 24776 facecrops missing from train videos and 7744 facecrops missing from val videos. If this is a known issue or you have faced it, can...

anikdatos

For resolving https://github.com/okankop/ASDNet/issues/5

anikdatos

https://github.com/okankop/ASDNet/blob/a16c5ede4f07278b25e1f61a10623bd34b69437b/data/slice_audio_tracks.py#L12 Because only labels 0 and 1 are filtered out, label 2 ("SPEAKING_NOT_AUDIBLE") audio isn't saved in slices. While reading the labels in core/dataset.py (line 28), all 2 labels are...

anikdatos

The command `./scripts/dowloads.sh` should be `./scripts/downloads.sh` with a "n" in downloads.

edgarcin95

Is possible to share your pre-train model?

1

Hi author, Could you publish your model which obtain the performance of the paper, Thanks!

Leeviber

bugdary

Overcautious