Installation is as simple as running:
pip install git+https://github.com/wiktorlazarski/edge-aware-face-autoencoder.git
Baseline model without edge-awareness trained on images with 256px resolution, VAE latent dimension equals to 512 and reconstruction loss weight equals to 100 000. Edges' weight was set to 1.
Edge-aware model with edges' weight set to 3. All other parameters were the same as in the baseline model.
Edge-aware model with edges' weight set to 10. All other parameters were the same as in the baseline model.
# Clone repo
git clone https://github.com/wiktorlazarski/edge-aware-face-autoencoder.git
# Go to repo directory
cd edge-aware-face-autoencoder
# (Optional) Create virtual environment
python -m venv venv
source ./venv/bin/activate
# Install project in editable mode
pip install -e .[dev]
# (Optional but recommended) Install pre-commit hooks to preserve code format consistency
pre-commit install
# Clone repo
git clone https://github.com/wiktorlazarski/edge-aware-face-autoencoder.git
# Go to repo directory
cd edge-aware-face-autoencoder
# Create and activate conda environment
conda env create -f ./conda_env.yml
conda activate face_autoencoder
# (Optional but recommended) Install pre-commit hooks to preserve code format consistency
pre-commit install