Hello！
I still have a question, for the vertex transformation of 2d-3d space, for two-dimensional joint node position information, how the network learns the position relationship between the joint nodes in three-dimensional space, such as the front and back relationship, when the hand is tilted forward, the finger is in front, and the wrist point is relatively backward. How did the network learn this?
Hello！ I still have a question, for the vertex transformation of 2d-3d space, for two-dimensional joint node position information, how the network learns the position relationship between the joint nodes in three-dimensional space, such as the front and back relationship, when the hand is tilted forward, the finger is in front, and the wrist point is relatively backward. How did the network learn this?