Depth prediction of view transform

[weixin0107]

Thanks for your amazing work! When I read the UVTR paper and source code, I found the depth distribution predicted by image feature is not supervised. This setting makes me a little confused, why doesn't supervise it by LiDAR depth?

Looking forward to your reply, thanks sincerely!