DAEMONS

Potsdam data set of eye movement on natural scenes

DAEMONS dataset [Schwetlick et al., 2024].

The DAEMONS paper presents the Potsdam dataset of eye movements on natural scenes, aimed at advancing research in visual cognition and machine learning. It introduces a large-scale dataset with 2,400 images and eye-tracking data from 250 participants, ensuring high-quality data collection using state-of-the-art equipment. The study focuses on both fixation distributions and scan paths, making the dataset valuable for various modeling approaches, including saliency prediction and cognitive modeling.

The dataset is split into train (precomputed_events[0]) and validation (precomputed_events[1]).

Check the respective paper for details [Schwetlick et al., 2024].

DatasetDefinition class implementation: DAEMONS