I noticed that SemiMixHead1 contains interfaces for losses such as GaussianOverlapLoss, GaussianVoronoiLoss, and EdgeLoss, but I could not find their corresponding implementations in the current repository.
Since I am still trying to understand the code while comparing it with the method described in the paper, I would like to ask whether these losses are part of the final method used in the paper, or whether they are interfaces left from earlier experiments.
I noticed that SemiMixHead1 contains interfaces for losses such as GaussianOverlapLoss, GaussianVoronoiLoss, and EdgeLoss, but I could not find their corresponding implementations in the current repository.
Since I am still trying to understand the code while comparing it with the method described in the paper, I would like to ask whether these losses are part of the final method used in the paper, or whether they are interfaces left from earlier experiments.