Treffer: Automation of the evaluation of complex eye-tracking data

Eye-tracking technology has become increasingly vital in various fields, such as psychology, marketing, and transportation, offering deep insights into visual attention and cognitive processes. One of the biggest challenges in the detection of areas of interest that are defined through special markers (ArUco) is the dynamic perspective changes. This project aims to improve and rewrite an existing system for automating the detection of gazes onto dynamic areas of interest using real-world data from train driver cabs. The enhanced system employs advanced algorithms in combination with ArUco markers to identify and locate dynamic areas of interest, such as the tachometer, windows or dashboards. This thesis focuses on automating the analysis of complex eye-tracking data to enhance efficiency, accuracy, and maintainability, specifically within the context of train driver cabs. Key advancements include integrating pose estimation and a modular design to improve areas of interest detection and system maintainability. The system utilizes Python and OpenCV libraries for marker detection and pose estimation, significantly enhancing accuracy and robustness. Additionally, the system adheres to clean code principles and is comprehensively documented to ensure usability and maintainability by non-expert users. The thesis demonstrates notable improvements in processing efficiency and areas of interest detection accuracy, resulting in a robust tool for analyzing eye-tracking data in train cab environments. The solution was validated using videos and eye-tracking data of real-world train driver cabs, showcasing the system’s capability to handle real-world scenarios effectively. ; Die Eye-Tracking-Technologie hat in verschiedenen Bereichen wie Psychologie, Marketing und Verkehrswesen zunehmend an Bedeutung gewonnen und bietet tiefe Einblicke in die visuelle Aufmerksamkeit und kognitive Prozesse. Diese Arbeit konzentriert sich auf die Automatisierung der Analyse komplexer Eye-Tracking-Daten, um die Effizienz, Genauigkeit und ...