Machine Learning, NLP

Self-Supervised Monocular 3D Face Reconstruction

MGCNet

Self-Supervised Monocular 3D Face Reconstruction by Occlusion-Aware Multi-view Geometry Consistency(ECCV 2020)

This is an official python implementation of MGCNet. This is the pre-print version https://arxiv.org/abs/2007.12494.

  1. video
    ECCV2020_Github

  2. image
    result_multiPose

  3. Full video can be seen in [YouTube] https://www.youtube.com/watch?v=DXzkO3OwlYQ

1. Code + Requirement + thirdlib

We run the code with python3.7, tensorflow 1.13

git clone --recursive https://github.com/jiaxiangshang/MGCNet.git
cd MGCNet
(sudo) pip install -r requirement.txt

(1) For render loss(reconstruction loss), we use the differential renderer named tf_mesh_render(thanks!) https://github.com/google/tf_mesh_renderer.
I find many issue happens here, so let’s make this more clear.
The tf_mesh_render does not return triangle id for each pixel after rasterise, we do this by our self and add these changes as submodule

 

 

 

To finish reading, please visit source site