"The method surpasses the performance of the baselines in the leading metric (F-Score)" mean that it outperforms all five baselines (PICO-MR, EVP++, Marigold, DepthAnythingv2, Garg) on the F-score, or just a specific one?
Posted by: al @ Feb. 1, 2025, 10:54 a.m.I think Marigold and DepthAnythingv2, cause they are the baseline, other 2 are not baseline, they are from previous challenge top solutions.
Posted by: lavreniuk @ Feb. 4, 2025, 4:32 p.m.Both of them? Then the baseline is F=18.64?
Posted by: al @ Feb. 4, 2025, 4:48 p.m.yes, I think so
Posted by: lavreniuk @ Feb. 4, 2025, 5:33 p.m.Sorry for the late reply. We consider as "baselines" Marigold, Depth Anything v2, and Garg's baseline.
Any of them should be beaten, at least in one metric.
Of course, the participants are encouraged to beat the winner from the previous edition :)
Posted by: mdec @ Feb. 26, 2025, 3:54 p.m.And just to clarify: to some extent, the method has to be "original", which means it is not simply a reproduction of an existing method/baseline without any personal contribution.
This will be assessed after the challenge is over when participants will be asked to produce a small report about their method