Abstract
Analysis of eye tracking data often requires accurate classification of eye movement events. Human experts and classification algorithms often confuse episodes of fixations and smooth pursuits because their feature characteristics overlap.
To support the development of better classification methods, we created a benchmark data set that does not rely on human annotation as the gold standard. It consists of almost four hours of eye movements from ten participants who fixated targets designed to induce saccades, fixations, and smooth pursuits under highly controlled conditions.
The paper makes both the raw data and a companion Python package available for preprocessing and assigning plausible benchmark labels. The goal is to support feature engineering and the training, validation, and benchmarking of eye-movement classification algorithms.