Data associated with the publication: EyePose: pose-guided saccadic eye movement video generation for deep learning-based neurologic disease phenotyping

The synthetic dataset comprises 3,000 labeled saccadic eye movement videos evenly distributed across three classes: normal, hypometria, and hypermetria. Each video simulates horizontal saccades with realistic amplitude and velocity patterns, and is annotated based on saccadic accuracy. The videos are sampled at 25 Hz with a resolution of 256×256 pixels, and span approximately 5 seconds, capturing a sequence of multiple saccadic events per clip. For each video, a corresponding waveform is provided.