Publication

Abstract : This paper presents a comparison of event detection task in eye movement with the exact events recorded from eye tracking device. The primary goal of this research work is to build a general approach for eye-movement based event detection, which will work with all eye tracking data collected using different eye tracking devices. It utilizes an end to end method based on deep learning, which can efficiently utilize eye tracking raw particulars that is further grouped into Saccades, post-saccadic oscillations and Fixations. The drawback of deep learning method is that it requires a lot of preprocessing data. At first, we have to build up a strategy to enlarge handcoded information, with the goal that we can unequivocally augment the informational index utilized for preparing, limiting the run through time on coding by a human. Utilizing this all-encompassing hand-coded information, we instruct neural networks model to process eye-development fixation grouping from eye-movement information in the absence of any previously defined extraction or post-preparing steps.