Closed forest520 closed 2 years ago
If you don't care about head motion, 2D landmarks are also an option. The benefit of 3D landmarks is to disentangle the expression motion, head motion, and camera matrix. Directly using 2D landmarks entangles these parameters.
Dear Dr. Lu: In the paper, sparse 3D landmark are used as an intermediate representation, and later project these 3D positions to the 2D image plane via pre-computed camera intrinsic parameters 𝐾 to get Conditional Feature Maps. I wonder whether it is ok to use 2D landmark directly if I don't need head motion as mentioned in #11 Thanks again for your answer!