cleardusk / 3DDFA_V2

The official PyTorch implementation of Towards Fast, Accurate and Stable 3D Dense Face Alignment, ECCV 2020.
MIT License
2.9k stars 511 forks source link

How to get "sparse" 3d facial landmarks #103

Open tkzky opened 3 years ago

tkzky commented 3 years ago

Thank you for your code. I'd like to know how can I get "sparse" 3D landmark coordinates? Like (x,y,z) And does the Z coordinate represent the depth information of the landmarks? Looking forward to your reply!

lewisandJiang commented 3 years ago

I also have this problem, do you solve it ?

tkzky commented 2 years ago

I don't solve it, what about you?

cleardusk commented 2 years ago

https://github.com/cleardusk/3DDFA_V2/blob/9fdbea1eb97f762221f71f5c76f08f52296c6704/TDDFA.py#L134

But the z is not the real depth, since 3DDFA_V2 adopts the weak-perspective projection (or scaled orthogonal projection).