Aryan Jain
Aryan Jain
I noticed that you guys have a torch.cuda.amp.autocast() in the _evaluate function ([line 143](https://github.com/hushell/pmf_cvpr22/blob/main/engine.py#L143)) in engine.py. In my experience, autocasting is typically used in conjunction with a GradScaler as well....
I want to track points across many short horizon videos of, say, dimension (8, 256, 256, 3). Suppose I have B such videos and I want to track the same...
Hello, the current demo datasets don't include any depth maps. How should I go about recreating the dataset so it has depth maps as an observation along with the rgb...