Road accidents are skyrocketing, and traffic safety is a severe problem around the world. Many road traffic deaths are related to drivers’ unsafe behaviors. In this paper, we propose two different deep-learning models which classify the driver’s actions in a 60-second time frame into two main categories: Normal and Aggressive driving based on GPS data collected at 1 Hz, which is later preprocessed and passed to the proposed models to identify dominant driving behavior in each time frame. The models achieved an accuracy of 93.75 percent in real-world tests, which proves the efficiency of this method in driving behavior recognition.
Driving Behaviors Recognition Using Deep Neural Networks
2023
9-12 Pages
Embedded Selforganising Systems, Vol. 10 No. 5 (2023): Applied AI Solutions on Edge Devices
Sonstige
Elektronische Ressource
Englisch
CAR TYPE RECOGNITION WITH DEEP NEURAL NETWORKS
British Library Conference Proceedings | 2016
|