top of page

Country

Mr. John Smith

Job title

Company

People

Distracted driving is one of the major causes of vehicle accidents. Accurate driver behaviour detection has the potential to reduce distracted driving and improve road safety. This paper introduces a methodology to use video data to detect driver behaviours with the support of a 'Deep Learning' Convolutional Neural Network (CNN) model. 11,970 video clips, recorded from three viewing angles, were used to train 3 independent models. The dataset contains eight classification targets, including concentration, distraction, and playing with phones. The study achieved accuracy rates of 94-100%, which varied by target class. The accuracy was also influenced by the location of the video camera, with the best model using video clips from a camera set on the participant's right side with its lens angled down towards the participant. This study demonstrates that camera angle can influence the quality of training data and establishes that deep learning models can be trained to accurately identify a wide range of distracted behaviours.



RMIT University: Kun Zou, M. Fard, J.L. Davy, S.R. Robinson

Detection of Distracted Driver Behaviour Using a Deep Learning Model: Influence of Camera Viewing Angle

APAC-21-169 • Paper

DOWNLOAD PAPER PDF
DOWNLOAD POSTER PDF
DOWNLOAD SLIDES PDF

Sign up or login to the ICC to download this item and access the entire FISITA library.

Upgrade your ICC subscription to access all Library items.

Congratulations! Your ICC subscription gives you complete access to the FISITA Library.

BUY NOW

Retrieving info...

Available for purchase on the FISITA Store

OR

bottom of page