Small Vehicle Dataset: Aerial and CCTV Imagery

This dataset comprises two curated subsets for small vehicle detection: Aerial Vehicle Dataset with 9,685 images (train/val/test split: 6,779/1,935/971) CCTV Vehicle Dataset with 11,109 images (train/val/test split: 7,776/2,223/1,110) Both subsets provide vehicle annotations in YOLO PyTorch format at a standardized resolution of 640×640 pixels. A single class label “vehicle” covers various vehicle types including cars, buses, motorcycles, and trucks. Designed to support research in challenging real-world scenarios, the datasets include diverse traffic densities, lighting conditions, and viewpoints. The data combines manually annotated images and samples sourced from established aerial vehicle datasets [Azimi et al., 2020; Lin et al., 2020; Wang et al.,2022; Wen et al., 2020].