Help me! on-prem deployed CVAT | Video annotation

[prabaljainn]

Hi I deployed my CVAT on my server and I did some annotation to one of my task which is a video 75secs, it takes a hell lot of time to export annotations for even a single video that too 7-8 gb zip..

My server specs are pretty good.
GPU with 128gbs RAM, i10 12th gen.

--> how to support this heavy task with GPU?
--> why it takes so much time? can we imrove?
--> should i consider trasforming my video to frames as 80% of the frames are unnesseary.

[azhavoro]

@prabaljainn Hello,

Unfortunately you haven't provided any data to compare exporting annotations with and without images, but I assume most of the time is spent decoding video and encoding png images. Of course, converting video to images should reduce the export time, because in this case the image is just copied to the destination archive.