HumanRecon

Official PyTorch implementation of the paper HumanRecon: Neural Reconstruction of Dynamic Human Using Geometric Cues and Physical Priors.

Datasets

Download the Human3.6M datase from here and ZJU-Mocap dataset from here.

Environment

python==3.7
torch==1.4.0+cu100

Traning

Take the training on S9 as an example.

python train_net.py --cfg_file configs/aninerf_s1p.yaml exp_name aninerf_s1p resume False

Testing

Take the testing on S9 as an example.

python run.py --type evaluate --cfg_file configs/aninerf_s9p.yaml exp_name aninerf_s9p resume True

Acknowledgments

Some components of this code implementation are referenced from NeuralBody and Animatable-NeRF. We sincerely appreciate their valuable contributions.