Why the representation model's input is 17 keypoints? Is it possible to pretrain a 26 keypoints( include the heel and foot big toe such as halpe) version? Because the angle of foot is important in some videos. or where the code should be changed if i want to pretrain a 26 keypoints or more version.
Why the representation model's input is 17 keypoints? Is it possible to pretrain a 26 keypoints( include the heel and foot big toe such as halpe) version? Because the angle of foot is important in some videos. or where the code should be changed if i want to pretrain a 26 keypoints or more version.