CMU-Perceptual-Computing-Lab / panoptic-toolbox

http://domedb.perception.cs.cmu.edu/
416 stars 80 forks source link

ground truth 3d annotation univTime #80

Closed SvitlanaKyrychenko closed 1 year ago

SvitlanaKyrychenko commented 1 year ago

I downloaded 171204_pose1 and want to compare my results with ground truth but for now do not understand how to correspond video frames with ground truth json files.

This is the first json file image

This is the last json file image

Could you explain how "univTime" is formed for each file? Also is annotation starts from the first second of a video?

Thanks in advance for the answer.