Andrew-Brown1/Automated_Face_Tracking_and_Labelling

Videos not written correctly

Closed this issue · 3 comments

When the code runs, only a small percentage of annotated videos is actually written.
For example when annotation 100 videos, annotations are returned for all 100, but only ~10 videos are written.

Hi,

So do I understand correctly that this is when using "--make_annotation_video True" ?

i.e. using "python VideoFaceAnnotator.py --make_annotation_video True"

Yes.

Also, just a note, this will always be True in your current implementation.
You'd probably want to change the bool flags to:

parser.add_argument('--store_video', default=False, action='store_true', ...)

Closed -- solved with the update of all packages which leads me to believe there was an ffmpeg issue somewhere which happened silently.