Model for viewing art

It is difficult for us to express internal sensitivities such as our art preferences with words. Recently, some psychological studies have reported a relationship between preference and eye movement. Several engineering studies have reported on methods or systems for tracking eye movement. In this study, we hypothesized that eye movement when viewing a painting can reflect preferences, and we tried to infer preferences by eye movement of participants while viewing paintings. If human preferences for art are modeled, the system can present a way to adapt to individual preferences. First, we measured the gaze points of participants as they viewed each painting with an eye-tracking system. Using the Bayesian method, we modeled the gaze points of participants as a probability distribution as they viewed a painting. We modeled their impressions while viewing paintings based on the gaze distribution.