GazeLei / gazeclr

PyTorch implementation for Contrastive Representation Learning for Gaze Estimation

Contrastive Representation Learning for Gaze Estimation

This repository is the official PyTorch implementation of GazeCLR.

• Published at NeurIPS 2022, Gaze Meets ML (Best Paper Award, Spotlight)
• Authors: Swati Jindal, Roberto Manduchi

Requirements

The code is tested with Python 3.7.10 and torch 1.18.1.

To install all the packages:

pip install -r requirements.txt

Data Processing

1. Download EVE Dataset
2. We extract all frames from videos in pre-processing stage for faster dataloading during training. Run

python video2imgs.py

Train GazeCLR

GazeCLR (Equiv)

python main.py --config_json configs/gazeclr.json --save_path <path/to/save> --is_load_label --same_person

GazeCLR (Inv+Equiv)

python main.py --config_json configs/gazeclr_inv_equiv.json --save_path <path/to/save> --is_load_label --same_person

Pre-trained Models

You can download pretrained models here:

• GazeCLR (Inv+Equiv)
• GazeCLR (Equiv)

Questions?

For any inquiries, please contact at [e-mail]