This is a great idea. Single-video pose estimation has just become decent hasn't it? I guess the problem is the "matching" of the two pose animations to give th

This is a great idea. Single-video pose estimation has just become decent hasn't it? I guess the problem is the "matching" of the two pose animations to give the minimum difference.