Enhancing Eye Ball Tracking System for Enhanced Human Computer Interaction: Deep Learning Base CNN

S. Suganya, K. Balamurugan, M. Sasipriya, R. Sowmiya, S. Krishna Veerandhra Suthir, S. Selvamani

2025

Abstract

AIM : This study aims to improve eye-tracking systems by using a hybrid model that combines Convolutional Neural Networks (CNNs) for semi-supervised anomaly detection. Materials and Method: The model was trained on a comprehensive multimodal dataset, including motion, eye-tracking inputs, and action context, to extract relevant features and identify complex patterns. Group 1: The existing method uses RNN for gaze analysis and Anomaly detection with 1000 to 10000 image samples. Group 2: The proposed model uses CNN as the same samples effectively learns from distributed multimodal datasets. Result: The reducing anomaly detection processing time from 57.5 ms to 52 ms, achieving a 94.5% confidence interval and an accuracy improvement of 9.5% over baseline models. The significance value of 0.0028 highlights the approach's efficacy in detecting gaze anomalies and optimizing system performance. Conclusion: In this work, it is observed that CNN has significantly better results than RNN.

Paper Citation

in Harvard Style

Suganya S., Balamurugan K., Sasipriya M., Sowmiya R., Suthir S. and Selvamani S. (2025). Enhancing Eye Ball Tracking System for Enhanced Human Computer Interaction: Deep Learning Base CNN. In Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies - ICRDICCT`25; ISBN 978-989-758-777-1, SciTePress, pages 264-270. DOI: 10.5220/0013911600004919

in Bibtex Style

@conference{icrdicct`2525,
author={S. Suganya and K. Balamurugan and M. Sasipriya and R. Sowmiya and S. Suthir and S. Selvamani},
title={Enhancing Eye Ball Tracking System for Enhanced Human Computer Interaction: Deep Learning Base CNN},
booktitle={Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies - ICRDICCT`25},
year={2025},
pages={264-270},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013911600004919},
isbn={978-989-758-777-1},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies - ICRDICCT`25
TI - Enhancing Eye Ball Tracking System for Enhanced Human Computer Interaction: Deep Learning Base CNN
SN - 978-989-758-777-1
AU - Suganya S.
AU - Balamurugan K.
AU - Sasipriya M.
AU - Sowmiya R.
AU - Suthir S.
AU - Selvamani S.
PY - 2025
SP - 264
EP - 270
DO - 10.5220/0013911600004919
PB - SciTePress