Presentation Title

Detection of Motor Vehicle Operator Attention from a Single Camera

Start Date

November 2016

End Date

November 2016

Location

MSE 113

Type of Presentation

Oral Talk

Abstract

According to the Centers for Disease Control and Prevention, motor vehicle accidents resulting in death caused $44 billion in medical and work loss in 2013. Distracted driving is the leading cause of motor vehicle accidents. Vehicles are equipped with an increasing number of safety features, yet an effective system to warn drivers when they are inattentive has yet to be revealed. Current attention monitoring systems include cameras to monitor gaze of the driver, and a system to monitor the gestures of the driver using multiple cameras placed around the car. There is no consistent method to measure attention from gaze, and multi-camera systems are difficult to deploy in the field. We propose a system to monitor the driver from a single camera using facial recognition technology. The system analyzes the driver in real time and uses physiological cues to determine if they are distracted. Experiments found that detecting the face of the driver in a video presents the greatest challenge to our method, and the focus of work has been the development of algorithms that are capable of detecting a face for our problem. Results show that the Constrained Local Models algorithm shows promise for the proposed system. While it is expected that autonomous cars will decrease motor vehicle accidents, they will revert control to the human operator in emergencies. This can only occur when the driver is safely paying attention. A system to monitor the attention of a driver will impact society greatly, and the need for such a system will not diminish in the future.

This document is currently not available here.

Share

COinS
 
Nov 12th, 11:30 AM Nov 12th, 11:45 AM

Detection of Motor Vehicle Operator Attention from a Single Camera

MSE 113

According to the Centers for Disease Control and Prevention, motor vehicle accidents resulting in death caused $44 billion in medical and work loss in 2013. Distracted driving is the leading cause of motor vehicle accidents. Vehicles are equipped with an increasing number of safety features, yet an effective system to warn drivers when they are inattentive has yet to be revealed. Current attention monitoring systems include cameras to monitor gaze of the driver, and a system to monitor the gestures of the driver using multiple cameras placed around the car. There is no consistent method to measure attention from gaze, and multi-camera systems are difficult to deploy in the field. We propose a system to monitor the driver from a single camera using facial recognition technology. The system analyzes the driver in real time and uses physiological cues to determine if they are distracted. Experiments found that detecting the face of the driver in a video presents the greatest challenge to our method, and the focus of work has been the development of algorithms that are capable of detecting a face for our problem. Results show that the Constrained Local Models algorithm shows promise for the proposed system. While it is expected that autonomous cars will decrease motor vehicle accidents, they will revert control to the human operator in emergencies. This can only occur when the driver is safely paying attention. A system to monitor the attention of a driver will impact society greatly, and the need for such a system will not diminish in the future.