Reconstruct 3D model from 2D human face images and CNN based PCA generation.
Environment:
PyTorch 1.5 CUDA 10.2
Extra packages:
Training Dataset: https://bcsiriuschen.github.io/CARC/
Face Alignment: https://github.com/1adrianb/face-alignment
Soft Rasterizer: https://github.com/ShichenLiu/SoftRas
Mesh Lab: https://www.meshlab.net/
This is the project from Yue Wu and Juheng Yang in Image and Video Processing Spring 2021.
Thanks a lot for the help from Zhipeng Fan.
Update: 7/24/2021 SoftRenderer is removed in the source code, but there are still some examples provided by the author on how to use SoftRasterizer.