01656420

Monograph

http://www.trb.org/Main/Blurbs/176774.aspx

The authors of this report have recently developed a novel computer vision based technology for distraction and fatigue detection that is based on a single CCD camera and uses deformable model technology to track from video the geometry of any person’s face, his/her expressions and measure slow eyelid closure which is directly related to fatigue. The system is non-intrusive and can work with both visible and infrared light regardless of the driver background which changes over time during driving. The system tracks the whole face and therefore can reliably detect the eyes and the eyelids. The system is also potentially cost-effective since it is based on an off-the-shelf computer and a single camera. The three primary objectives of this research project were: 1.) Develop a prototype of driver facial monitoring system via porting the software onto a portable processor to create a hardware system; 2.) Initial prototype testing using existing or acquired videos to test the accuracy of the tracking ability of the prototype with a variety of facial features, including glasses, different skin tones, facial hair; with different levels of illumination; and with various degrees of facial movement; and 3.) Public beta testing of prototype on buses to evaluate the accuracy in terms of: (a.) validity –that the system accurately measures what it is supposed to measure, and (b.) sensitivity – that the system picks up distraction and fatigue with minimal false positives and negatives.

This Transit-IDEA investigation was conducted by Rutgers University.

Transit IDEA Project 80

English

23p

2017-9

TRANSIT-IDEA Program Project Final Report

Final Report

Digital/other

Figures; Photos; References; Tables

Bus drivers; Computer vision; Distraction; Eye movements; Face; Fatigue (Physiological condition); Prototype tests; Prototypes; Transit operators; Video

Innovations Deserving Exploratory Analysis (IDEA) (Program)

Motor Carriers; Public Transportation; Safety and Human Factors

TRIS, TRB, ATRI

Dec 21 2017 2:23PM