Real-Time Driver State Monitoring Using a CNN Based Spatio-Temporal\n Approach

2019/07/18 by Neslihan Köse, Kose, Neslihan, Okan Köpüklü +5
Engineering · Computer Science · #Autonomous Vehicle Technology and Safety #Video Surveillance and Tracking Methods #Anomaly Detection Techniques and Applications

paper · pdf · doi:10.48550/arxiv.1907.08009

Abstract

Many road accidents occur due to distracted drivers. Today, driver monitoring\nis essential even for the latest autonomous vehicles to alert distracted\ndrivers in order to take over control of the vehicle in case of emergency. In\nthis paper, a spatio-temporal approach is applied to classify drivers'\ndistraction level and movement decisions using convolutional neural networks\n(CNNs). We approach this problem as action recognition to benefit from temporal\ninformation in addition to spatial information. Our approach relies on features\nextracted from sparsely selected frames of an action using a pre-trained\nBN-Inception network. Experiments show that our approach outperforms the\nstate-of-the art results on the Distracted Driver Dataset (96.31%), with an\naccuracy of 99.10% for 10-class classification while providing real-time\nperformance. We also analyzed the impact of fusion using RGB and optical flow\nmodalities with a very recent data level fusion strategy. The results on the\nDistracted Driver and Brain4Cars datasets show that fusion of these modalities\nfurther increases the accuracy.\n

Related

• Enhancing Road Safety: Real-Time Detection of Driver Distraction through Convolutional Neural Networks
• Spatial-temporal Fusion Convolutional Neural Network for Simulated Driving Behavior Recognition
• Keep Your AI-es on the Road: Tackling Distracted Driver Detection with Convolutional Neural Networks and Targeted Data Augmentation
• Detection of Distracted Driver using Convolution Neural Network
• Visual Dominance and Emerging Multimodal Approaches in Distracted Driving Detection: A Review of Machine Learning Techniques