Using gaze for behavioural biometrics

Identification system based on eye movements modeled as an Ornstein-Uhlenbeck process (Paper).

Installation

Install required libraries with Anaconda:

conda create --name gazeID -c conda-forge --file requirements.txt
conda activate gazeID

Install NSLR-HMM

python -m pip install git+https://gitlab.com/nslr/nslr

Features extraction

Extract Ornstein-Uhlenbeck features from FIFA-DB dataset (datasets/FIFA) launching the module extract_OU_params.py, results will be saved in features/FIFA_OU_posterior_VI.

Train and test

Module event_kfold.py exploits SVMs for classification on the features extracted as an Ornstein-Uhlenbeck process via a Nested cross-validation procedure.

Reference

If you use this code, please cite the paper:

@article{damelio2023using,
AUTHOR = {D’Amelio, Alessandro and Patania, Sabrina and Bursic, Sathya and Cuculo, Vittorio and Boccignone, Giuseppe},
TITLE = {Using Gaze for Behavioural Biometrics},
JOURNAL = {Sensors},
VOLUME = {23},
YEAR = {2023},
NUMBER = {3},
ARTICLE-NUMBER = {1262},
URL = {https://www.mdpi.com/1424-8220/23/3/1262},
ISSN = {1424-8220},
DOI = {10.3390/s23031262}
}

@article{d2021gazing,
  title={Gazing at Social Interactions Between Foraging and Decision Theory},
  author={D'Amelio, Alessandro and Boccignone, Giuseppe},
  journal={Frontiers in neurorobotics},
  volume={15},
  pages={31},
  year={2021},
  publisher={Frontiers}
}

@article{boccignone2020gaze,
  title={On gaze deployment to audio-visual cues of social interactions},
  author={Boccignone, Giuseppe and Cuculo, Vittorio and D’Amelio, Alessandro and Grossi, Giuliano and Lanzarotti, Raffaella},
  journal={IEEE Access},
  volume={8},
  pages={161630--161654},
  year={2020},
  publisher={IEEE}
}