N-UCF Sports: A Neuromorphic version of the UCF Sports Saliency Dataset

N-UCF Sports is a synthetic event-camera dataset for video saliency prediction, derived from the UCF Sports action dataset. Event streams were generated from the original RGB video sequences using the ESIM event-camera simulator, producing per-pixel brightness-change events that approximate the output of a real event sensor. Each sequence is paired with the corresponding ground-truth saliency maps, enabling training and evaluation of saliency models directly in the event domain.

The dataset is intended for research in event-based vision and neuromorphic sensing, particularly dynamic saliency prediction, and  attention modelling.