3D人脸网格识别

# 3D Face recognition based on Geometric Deep Learning This is the repository for the preliminary master thesis. This project is a proof of concept method for a new type of 3D facial recognition. ## Setup instructions ``` conda create -n pytorch3d_new python=3.8 conda activate pytorch3d_new conda install -c pytorch pytorch=1.7.0 torchvision cudatoolkit=11.0 conda install -c conda-forge -c fvcore fvcore conda install -c pytorch3d-nightly pytorch3d ``` This repo also uses the dataset FaceGen, which is not publicly available at the moment. The root-dir in facegenDataset must point to the dataset folder. ## Running The project is tested with Cuda 11.0 To run it, do `export CUBLAS_WORKSPACE_CONFIG=:4096:8; python main.py` The export statement is to set reproducibility in the nvidia cublas api, and will not work without. Ref https://docs.nvidia.com/cuda/cublas/index.html#cublasApi_reproducibility