An Eye Gaze Real Time Road Detection System for Driver Assistance

Automated estimation of the allocation of a driver’s visual attention may be a critical component of future Advanced Driver Assistance Systems. In theory, vision-based tracking of the eye can provide a good estimate of gaze location. In practice, eye tracking from video is challenging because of sunglasses, eyeglass reflections, lighting conditions, occlusions, motion blur, and other factors. Estimation of head pose, on the other hand, is robust to many of these effects, but cannot provide as fine-grained of a resolution in localizing the gaze. However, for the purpose of keeping the driver safe, it is sufficient to partition gaze into regions. In this effort, we propose a system that extracts facial features and classifies their spatial configuration into six regions in realtime. Our proposed method achieves an average accuracy of 91.4% at an average decision rate of 11 Hz on a dataset of 50 drivers from an on-road study. IntroductionDriver distractions are the leading cause of mostvehicle crashes and near-crashes. According to a studyreleased by the National Highway Traffic SafetyAdministration (NHTSA) and the Virginia TechTransportation Institute (VTTI) 80% of crashes and65% of near-crashes involve some form of driverdistraction. In addition, distractions typically occurredwithin three seconds before the vehicle crash. Recentreports have shown that from 2011 to 2012, thenumber of people injured in vehicle crashes related todistracted driving has increased 9%. In 2012 alone,3328 people were killed due to distracted drivingcrashes, which is a slight reduction from the 3360 in2011.Distracted driving is defined as any activity that coulddivert a person’s attention away from the primary taskof driving. Distractions include texting, using a smartphone, eating and drinking, adjusting a CD player,operating a GPS system or talking to passengers. This is particularly challenging nowadays, where awide spectrum of technologies have been introducedinto the car environment. Consequently, the cognitiveload caused by secondary tasks that drivers have tomanage has increased over the years, hence increasingdistracted driving. According to a survey, performing ahigh cognitive load task while driving affects drivervisual behavior and driving performance. Referencesreported that drivers under high cognitive loadsshowed a reduction in the time spent examiningmirrors, instruments, traffic signals, and areas aroundintersections. Especially concerning is the use of hand-held phones and other similar devices while driving.NHTSA has reported that texting, browsing, anddialing cause the longest period of drivers taking theirEyes Off the Road (EOR) and increase the risk ofcrashing by three fold. A recent study shows that thesedangerous behaviors are wide-spread among drivers,54% of motor vehicle drivers usually have a cell phonein their vehicles or carry cell phones when they drive.Result in drastic changes in the facial features (e.g.,pupil and eye corners) to be tracked the system mustbe accurate for a variety of people across multipleethnicities, genders, and age ranges. Monitoring driver activities forms the basis of a safetysystem that can potentially reduce the number ofcrashes by detecting anomalous situations. Authorsshowed that a successful vision-based distracted