PyTorch implementation of the paper "Improving 2D Face Recognition via Discriminative Face Depth Estimation", ICB 2018 https://ieeexplore.ieee.org/document/8411215
I needed to train the model on two datasets BIWI (http://robotics.dei.unipd.it/reid/index.php/8-dataset/2-overview-biwi) and Pandora (http://imagelab.ing.unimore.it/pandora/) The file names refer to images coming from these two datasets. By following a similar approach, you can adapt it to the datasets you wish.