palver7
About the results of StdConvs vs EquiConvs
From the results of README and the paper, I find that the network constructed by Equirectangular Convolutions (EquiConvs) has no great improvement in indicators compared with the network constructed by traditional convolution (StdConvs). How can you explain the benefits of EquiConvs in processing ERP projection images?
I cannot explain the benefits myself. I can only point you to the part of the paper on robustness analysis. It appears EquiConvs is more robust against distortion of the ERP images caused by translation and rotation compared to StdConvs.
@palver7 Thanks, I tried it. But I find that implementing EquiConvs is time-consuming. The computation time of the network constructed by EquiConvs is 10 times that of the network constructed by StdConvs. So compared with EquiConvs, I prefer to choose StdConvs to build network.