Eye Tracking Algorithm based on Pyramid Convolutional Neural Network

Machine vision and pattern recognition approaches have achieved many successes in the field of visual recognition. One of the similar parts in this field is the detection of the eye area along with the tracking of the iris. As the second decade of research in eye movement biometrics begins, the previously demonstrated results strongly support the promising prospects of this field. The human eye system is responsible for coordinating eye movements to filter the influx of information from the visual environment. The problem of detecting, tracking and tracing the iris is a hot research topic in the field of biometrics and artificial intelligence sciences and under the set of machine vision, digital image processing and statistical and analytical pattern recognition. During the last few years, various methods based on smart structures, mechanical methods, electrical methods and optical methods (based on machine vision systems) have been active in this field. In recent years, especially from ۲۰۰۷ to ۲۰۱۹, eye movement and iris tracking based on machine vision has become a dominant method in several fields, including cognitive science. But considering the weaknesses in the field of tracking and tracing the iris, including the use of points to identify the iris area in the eye, the failure to obtain optimal results in terms of evaluation criteria, especially when studying and not studying the iris with accuracy criteria and Calling, high computational complexity and high execution time, this issue is considered as a hot research area to provide more optimal methods. Therefore, based on the presented approach based on the machine vision principles and deep learning principles using the pyramid convolutional neural network, an attempt will be made to track the iris from the images in the videos.

کلیدواژه ها:

Eye Tracking ، Iris Detection ، Machine Vision ، Deep Learning