Chalearn 3D High-Fidelity Mask Face Presentation Attack Detection Challenge@ICCV2021 Forum

Go back to competition Back to thread list Post in this thread

> Track level aggregation

we noticed that in val set the results on leaderboard are calculated based on tracks, not on individual frames. And the score for track is selected as an average for the track.
Please tell, whether
1. Results on test set are also will be calulated on track level?
2. Can we use this information and make a track level classification model? Or we can't use the knowledge that 10 consequent images belong to the same track?


Posted by: oleggrinch @ June 11, 2021, 2:43 p.m.

We're curious about this too

Posted by: cyberlink_rdme @ June 11, 2021, 5:10 p.m.
Post in this thread