Title: An AI-driven support system for detecting distracted driving behaviour and improving safe driving

Authors: Makoto Ikeda; Naoki Tanaka; Hibiki Tanaka; Leonard Barolli

Addresses: Department of Information and Communication Engineering, Fukuoka Institute of Technology, Fukuoka, Japan ' Graduate School of Engineering, Fukuoka Institute of Technology, Fukuoka, Japan ' Graduate School of Engineering, Fukuoka Institute of Technology, Fukuoka, Japan ' Department of Information and Communication Engineering, Fukuoka Institute of Technology, Fukuoka, Japan

Abstract: Accidents often result from diminishing driving skills and reduced attention, a problem that is becoming more prevalent with the aging population. Moreover, the rise in infotainment functions in vehicles poses another challenge, further distracting drivers from their primary task. Recognising the gravity of this situation, this paper presents an advanced intelligent driving support system to identify distracted driving behaviours. The core of our detection mechanism is based on YOLOv5m algorithm, which is fine-tuned with hyperparameter optimisation. The proposed system is able to recognise various distracted driving behaviours by analysing the driver's hand movements. By evaluating the proposed system, we observed that the use of a batch size of 8 and conducting 200 epochs produced the highest classification accuracy. The detecting hand manipulations of In-Vehicle Infotainment (IVI) system were identified with 100% accuracy. Also, cell phone use was detected with a probability exceeding 90%. Moreover, the system successfully can forecast the distracted driving behaviour when the vehicle is moving.

Keywords: distracted driving detection; classification; hyperparameter optimisation; YOLO.

DOI: 10.1504/IJGUC.2025.145178

International Journal of Grid and Utility Computing, 2025 Vol.16 No.2, pp.183 - 190

Received: 30 Oct 2023
Accepted: 01 Mar 2024
Published online: 24 Mar 2025 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article