Master of Science
Bauer, Michael A.
A driver’s behavior is one of the most significant factors in Advance Driver Assistance Systems. One area that has received little study is just how observant drivers are in seeing and recognizing traffic signs.
In this contribution, we present a system considering the location where a driver is looking (points of gaze) as a factor to determine that whether the driver has seen a sign. Our system detects and classifies traffic signs inside the driver’s attentional visual field to identify whether the driver has seen the traffic signs or not. Based on the results obtained from this stage which provides quantitative information, our system is able to determine how observant of traffic signs that drivers are. We take advantage of the combination of Maximally Stable Extremal Regions algorithm and Color information in addition to a binary linear Support Vector Machine classifier and Histogram of Oriented Gradients as features detector for detection. In classification stage, we use a multi class Support Vector Machine for classifier also Histogram of Oriented Gradients for features. In addition to the detection and recognition of traffic signs, our system is capable of determining if the sign is inside the attentional visual field of the drivers. It means the driver has kept his gaze on traffic signs and sees the sign, while if the sign is not inside this area, the driver did not look at the sign and sign has been missed.
Shabani, Shabham, "Assessment of Driver's Attention to Traffic Signs through Analysis of Gaze and Driving Sequences" (2019). Electronic Thesis and Dissertation Repository. 6209.