本篇論文使用類神經網路結合本論文所提出的 Inner Corner-Pupil Center Vector (ICPCV)特徵，製作出可以讓使用者不用固定頭部而且不需要使用昂貴的硬體就能使用的眼動追蹤系統，並且與使用係數矩陣注視點估測演算法作比較，有更好的表現。;The eye is an important organ that human receives information via from the outside world. It also can convey information such as the direction of gaze, the brightness of the surrounding environment and express emotions. Therefore, eye tracking has always been a popular research topic.
In recent years, there are more and more eye trackers produced and on sale. The application of eye tracking is more wide-ranging such as psychology, medicine, education, virtual reality. But most of the eye tracker in the market is very expensive, and the head needs to be fixed in order to accurately estimate the direction of gaze.
This study develops an eye tracking system based on neural network and Inner Corner-Pupil Center Vector (ICPCV) feature which defined by ourselves. Make it allows user move his/her head and needn’t expensive hardware. We compared with coefficient matrix gaze estimation algorithms and better than it.