Open
Description
Hi, I have really a really strong 3D track of a FLAME model on some footage.
https://www.linkedin.com/posts/ericlevy_were-doing-rd-on-a-new-machine-learning-activity-7262912859997716481-D4Do?utm_source=share&utm_medium=member_desktop
How can I derive the 2D landmarks from this like how SPIGA does, where they follow the contour of the face?
Could I use the pose data from the other repo to replace the pose data that SPIGA comes up with, and get a better result?
Metadata
Metadata
Assignees
Labels
No labels