چكيده به لاتين
Abstract:
Eye tracking and eye-gaze estimation have been extensively investigated by machine vision researchers and psychology community over the past few decades. Still, because of the uniqueness of the eye, the variation in the size, scale, location, and different lighting conditions remains to be seen as a challenging problem. Eye tracking in the fields of neuroscience, psychology, and interaction between man and computer are many uses. Estimates to look at in the analysis of human attention, human factors in industrial engineering, marketing and advertising, look-up interface and car care systems play a vital role.
In the past, early estimation systems required hardware such as specialized lenses or skin-mounted electrodes, which, in addition to costing, had a very difficult installation process. These factors made the technology inefficient in any application. At the moment, eye tracking systems and eye-gaze estimation systems are using additional hardware such as infrared sensors.
In this thesis, a low-cost system with a deep learning approach has been implemented. The purpose of the proposed system is to be widely used first. It is also resistant to the appearance of the eyes of the people, the variation in ambient light, various head conditions and background images.
Keywords: Eye tracking, Estimation of vision, Machine vision, Deep learning
