Open ChiefGodMan opened 5 years ago
As the following image showed, which is 300vw/035/vid.avi frames, the green landmarks is groundtruth, and the white landmarks coming from generated vertices, which should be 3d when the left face boundary is invisible, but it just capture the boundary. Why? Anything have I missed?
I also found this problem. In this way, there will be a big error when fitting the side face.
As the following image showed, which is 300vw/035/vid.avi frames, the green landmarks is groundtruth, and the white landmarks coming from generated vertices, which should be 3d when the left face boundary is invisible, but it just capture the boundary. Why? Anything have I missed?