Enhanced Attention-Based ResNet for Driver Distraction Detection

Premalatha S., S. Jayachitra, Latha B., Santhosh M., Geethambari M., Hari Arulkumar

2025

Abstract

Aim: The research seeks to identify some common drivers' distractions such as texting and eating, looking aside, by designing an EAB ResNet model for analyzing the face, eye, and hand movements of a person. Materials and Methods: The image data is preprocessed through resizing normalization, and augmentation. It is fine-tuned with transfer learning on a large annotated dataset to detect behaviors like phone use and eating. Group 1: The model was developed using a Convolutional Neural Network method is accurate in 82% of dealing with the AUC dataset. Crucial elements were accomplished through the use of the Grad-CAM. Group 2: The suggested system is an essential tool for enhancing road safety since it can achieve high accuracy even in intricate and dynamic driving situations by using the ResNet architecture. Result: ResNet obtains values of precision above 96% with small variations, high precision ensures that very few normal driving actions. The ResNet model continuously attains better accuracy, ranging from 90.5% to 96.01%. Statistical Analysis is also done to ensure that the model works robustly as the mean accuracy returned 95%, the Standard deviation is 0.92, and the Standard mean Error is 0.239. Conclusion: The Enhanced ResNet achieves accuracy and precision is outperforming conventional CNNs. Improvements in the coming future include a mental distraction ability and a rise in real-world adaptability.

Paper Citation

in Harvard Style

S. P., Jayachitra S., B. L., M. S., M. G. and Arulkumar H. (2025). Enhanced Attention-Based ResNet for Driver Distraction Detection. 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 213-218. DOI: 10.5220/0013925400004919

in Bibtex Style

@conference{icrdicct`2525,
author={Premalatha S. and S. Jayachitra and Latha B. and Santhosh M. and Geethambari M. and Hari Arulkumar},
title={Enhanced Attention-Based ResNet for Driver Distraction Detection},
booktitle={Proceedings of the 1st International Conference on Research and Development in Information, Communication, and Computing Technologies - ICRDICCT`25},
year={2025},
pages={213-218},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013925400004919},
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 - Enhanced Attention-Based ResNet for Driver Distraction Detection
SN - 978-989-758-777-1
AU - S. P.
AU - Jayachitra S.
AU - B. L.
AU - M. S.
AU - M. G.
AU - Arulkumar H.
PY - 2025
SP - 213
EP - 218
DO - 10.5220/0013925400004919
PB - SciTePress