This repository contains the code for a PyTorch reproduction of the Cross-Encoder from the paper Cross-Encoder for "Unsupervised Gaze Representation Learning", by Sun et al.
Read the blog post for more information.
To install this project's dependencies run
pip install -r requirements.txt
Then, save the Columbia Gaze dataset to a data
folder. To preprocess the dataset and start training the Cross-Encoder run
python train_cross_encoder.py