An ai-driven computational model for evaluating object memorability

Abstract

Visual perception is a key element in architectural design processes that determines the impact of object placements and spatial arrangements on users. Many factors influence visual perception, including size, texture, form and shape, light and shadow, color, and semantics. Recall is a state related to perception, being aware of an object. Studies have observed that objects that people forget and remember are common. The aim of this thesis is to examine the relationship between memorability and object size and location using a computational model, taking into account the perceiver's changing perspective. As part of this thesis, comprehensive research has been conducted on computational visual perception approaches. Then, a computational visual perception model was developed. A computational method is presented to examine the relationship between size and memorability in the first stage, and the relationship between object location and memorability in the second stage. A case study was conducted to implement the developed model with the permission of the Istanbul Technical University Social and Human Sciences Scientific Research and Publication Ethics Committee. The data collected in the case study and the outputs produced by the model were analyzed. Pearson and Spearman correlation analyses and simple linear regression analyses were conducted between the mean scores obtained from the case study and the size and location-based memorability scores obtained from the model. These analyses revealed a strong, positive, and significant relationship between object size and participant data, while no significant relationship was observed between object location and memorability.