Marek A. Pedziwiatr

The paper shows that eye movements are guided by both image features and knowledge-driven object representations, with gaze patterns changing when objects are perceived as coherent.

The paper shows that semantically inconsistent objects attract more fixations than consistent ones, but contextualized meaning maps fail to capture this relationship.

This study compares meaning maps and saliency models in predicting human fixations, finding that deep neural networks outperform meaning maps in accuracy and responsiveness to meaning changes.