محمدرضا اسلاميان كوپائي

Neuro-physiological methods provide an interesting way to observe users as they interact with information systems, one popular method of which is eye tracking . This method records the user's eye movement patterns while checking information on the screen. Eye tracking is used to eva‎luate and measure the connection between the content of the screen and the user. Also, this method has a series of clear and distinct advantages, the most important of which is non- invasiveness, in which the user does not need even the slightest effort. In this way, even when users don't click on the mouse or type anything to search, they help us to provide continuous and continuous data in a more natural environment just by viewing the screen. In this research, we want to analyze eye tracking data by means of deep learning approaches, and for this purpose we will use deep neural networks to classify fixations in different modes. For this purpose, we use the eye data of 10 women and 7 men who watched 79 emotional images. The images shown to the subjects have three basic dimensions of feeling, i.e. pleasantness, arousal and dominance. Data classification was done by transfer learning and using Xception, InceptionV3, ResNet50V2, ResNet50, DenseNet201 and EfficientNetB4 deep neural networks whose weights were trained on the ImageNet dataset. The obtained results show that deep learning algorithms and especially deep neural networks are powerful and accurate tools for processing people's eye fixations points.

رديابي چشم , تثبيت , يادگيري عميق , شبكه هاي عصبي عميق

Eye tracking , fixation , deep learning , deep neural networks

Mohammadreza Eslamian Koupaei

Dr. Mohammadreza Daliri