Such a great work! But I noticed that the leg of the SMPL has some offset or shift comparing to the image sample's, is it normal?
Also, when I use some yoga picture as input, the output SMPL is unnatural and totally wrong. Is there any mechanism that can anomaly detection and filter out unnormal frame, to enhance the performance?
Such a great work! But I noticed that the leg of the SMPL has some offset or shift comparing to the image sample's, is it normal? Also, when I use some yoga picture as input, the output SMPL is unnatural and totally wrong. Is there any mechanism that can anomaly detection and filter out unnormal frame, to enhance the performance?
