3D-FaceReconstruction

1. Setup

1.1. Basel Face Model 2019

Obtain the Basel Face Model from these links:

• https://faces.dmi.unibas.ch/bfm/bfm2019/restricted/model2019_fullHead.h5 (full head model, the statistics for the back of the head based on little data and heuristics)
• https://faces.dmi.unibas.ch/bfm/bfm2019/restricted/model2019_bfm.h5 (model masked similar to the original Basel Face Model mask)
• https://faces.dmi.unibas.ch/bfm/bfm2019/restricted/model2019_face12.h5 (model cropped to the face region)

Put the model files into a data folder and update the paths in env.py accordingly.

1.2 Facial Keypoint detection

You can download the pretrained facial keypoint detector here: http://dlib.net/files/shape_predictor_68_face_landmarks.dat.bz2

1.3 Environment variables

Change the DATA_PATH and MODEL_PATH environment variables in env.py to point to the locations where you stored the above files. If you commit changes, ensure to temporarily stash env.py in order to not push your local paths to the repository.