This repository contains the implementation of the losses used in the paper "Hyperparameter-Free Losses for Model-Based Monocular Reconstruction".
We also provide the 3D annotations in MICC and FaceWarehouse datasets that were used for the evaluation of the models in the paper.
To install:
pip install -r requirements.txt
After that, you can run the tests to ensure everything is working properly
python -m unittest
Running an example on how to use the losses:
python main.py