Hello,I wonder can omnire get the semantic 2D information or map?

[zzz5y]

Thank you for your great work,I wonder omnire whether could do this, it seems that you didn't mention it in project page,thank you a lot for you great work!

[MMMaverick]

I tried to get a BEV view on NuScenes, but it goes out of the field of view. So I set a tilted perspective, but the result was very distorted. This might be an inherent issue with 3dgs.

[zzz5y]

I tried to get a BEV view on NuScenes, but it goes out of the field of view. So I set a tilted perspective, but the result was very distorted. This might be an inherent issue with 3dgs.

I think so too.

[KMnO4AndWaston]

Hello bro, have you ever found the solution to make omniRe rendering semantic?

[zzz5y]

Hello bro, have you ever found the solution to make omniRe rendering semantic?

I think i maybe need to take segmantic map from dataset.And then for lidar point , we need to project the lidar point to 2D image to find the points' segmantic. And then we can train like the means or shs_dc shs_rest and so on ,but it still a rough way，because about finding the segmantic of the point ,it needs us to transfrom the label into other ways,and after training,there are somes ways like softmax in 3D or 2D.
But i don't start to hanlde and try it.And maybe my idea is wrong!But i hopes it can help you