DADA-seg is a pixel-level annotated accident dataset. This dataset extracts dynamic background from event-based data to enhance static RGB images with higher temporal resolution. The dataset contains various key scenes obtained from traffic accidents and can be used for semantic segmentation.