The essential elements of VISTA-CenterPoint

[QingXIA233]

Hello, really impressive works! I see that the code in this repo is about VISTA-OHS, but what catches my eyes is your plug-and-play VISTA module applied in CenterPoint. I'm working on reproducing VISTA-CenterPoint mentioned in your paper. I add RV feature map and the cross-view attention neck, the attention loss. I wonder whether I should perform Regression and classification decoupling as well. Could you please give me some hints about how to implement VISTA-CenterPoint on my own since there is not much information about this on your paper. Thanks in advance.