reshow / PRNet-PyTorch

The training and evaluation code for PRNet (《Joint 3D Face Reconstruction and Dense Alignment with Position Map Regression Network》)
73 stars 16 forks source link

Quantitative comparison with methods which have different number of vertices! #12

Open HOMGH opened 4 years ago

HOMGH commented 4 years ago

Thanks for your great work! When evaluating 3D face reconstruction , The ECCV paper compared with 3DDFA which is based on BFM. The difference between 3DDFA(50k) and PRNet (40k)is the number of vertex, when I use 3DDFA, how can I compare with PRN in AFLW2000? Look forward to your reply!