"Motorcycle Accident Detection and Prediction"

"The Motorcycle Accident Detection and Prediction (MADP) is a novel open-access datasetvof images for detecting traffic accidents involving motorcycles. MADP is constructed from previously collected data, annotated to include exclusively videos from closed-circuit television cameras, classified into three categories: no accident, accident without a motorcycle, and accident involving a motorcycle. In total, the dataset consists of 2,261 video segments, from which individual frames are extracted and partitioned into training, testing, and validation subsets. The corresponding annotations are provided, as well as the full dataset in a memory-map format and separate versions for each subset."