Electronic Thesis and Dissertation Repository

Thesis Format



Master of Science


Computer Science


Bauer, Michael A.


Several studies have considered driver’s attention for a multitude of distinct purposes, ranging from the analysis of a driver’s gaze and perception, to possible use in Advanced Driving Assistance Systems (ADAS). These works typically rely on simple definitions of what it means to “see,” considering a driver gazing upon an object for a single frame as being seen. In this work, we bolster this definition by introducing the concept of time. We consider a definition of ”seen” which requires an object to be gazed upon for a set length of time, or frames, before it can be considered as seen by the driver. This is done by examining consecutive frames to find those where the driver’s gaze remains uninterrupted within a constant bounding box of a given traffic object over a series of frames. A time-considering approach to defining traffic objects as seen or unseen provides a more thoughtful and accurate measure of driver’s perception, as we avoid the naive assumption that gazing upon an object for a single frame is enough time for a driver to process the object gazed upon, which ultimately could prove vital to a wide array of ADAS and i-ADAS systems.

Summary for Lay Audience

This thesis introduces a novel approach for examining the notion of what a driver “sees” in the context of Advanced Driving Assistance Systems (ADAS) where determining what a driver sees is based on images (frames) from cameras and computation of a driver’s gaze. Previous research often assumes that what a driver “sees” is based on determining a driver’s gaze on a single frame. “Seeing”, however, is complex and is based on function of the human eye and human cognition and a computational approach that a single frame is sufficient is likely too limited. This research considers a driver’s gaze and objects across a number of frames. It looks at the impacts that adjusting the number of frames considered under a driver’s gaze can have on the number of objects seen. This is investigated through the utilization of various frame length thresholds as the basis for our definition of seen, allowing us to compare these thresholds. The work aims to provide conclusions on the impact of these varied frame length thresholds and whether future work on determining what a driver “sees” as part of ADAS would benefit from a more thoughtful definition of “seen”.

Creative Commons License

Creative Commons Attribution-Noncommercial 4.0 License
This work is licensed under a Creative Commons Attribution-Noncommercial-Share Alike 4.0 License.