Advancing Human-Computer Interaction: A Deep Learning Approach to Eye Tracking

Abstract

The accelerated progress of human-computer interface (HCI) technologies has generated considerable interest in creating efficient and accurate eye-tracking algorithms. This paper introduces a novel approach utilizing deep learning techniques for real-time eye tracking. Our study's proposed Convolutional Neural Network (CNN) trains on an Eye-Chimera database. The results were notable, as the model exhibited a commendable average accuracy rate of 97.67%. The algorithm under consideration underwent a comprehensive evaluation, which yielded data suggesting a meager error rate of 0.023. This result serves as evidence of the system's resilience and reliability. Moreover, this technology's commendable precision and minimal margin of error make it an invaluable instrument for augmenting user experiences and facilitating accessibility within virtual reality, gaming, and assistive technologies. Consequently, its potential for widespread implementation holds promise for significantly increasing many human-computer interaction applications, yielding good outcomes.