Use mediapipe to detect changes in facial expressions in the video and output a JSON file of ARKit Blendshapes.
conda create --name video2face
conda activate video2face
pip install -r requirements.txt
python run.py --input ./videos/ --output ./outputs/