User embeddings to predict personalized gaze patterns and scanpaths

Niveau : Master 2

Période : spring/summer 2025

Context
Understanding and predicting human gaze behavior has applications in various domains, including human-computer interaction, virtual reality, and cognitive research. Most existing models focus on general gaze paths, often guided by visual saliency, which assumes that visual attention is primarily driven by image features. However, real-world gaze behavior is highly idiosyncratic, varying significantly across individuals due to differences in interests, tasks, and cognitive processes [1,2]. By capturing and modeling these individual differences, we can create more realistic simulations of human attention and improve personalized interfaces and experiences [3].

Contact : alexandre.bruckert@univ-nantes.fr