Explore high-quality datasets for your AI and machine learning projects.
The FGVC‑Aircraft dataset is a benchmark for fine‑grained visual classification of aircraft, containing 10,200 images with 100 images per aircraft model, covering 102 distinct aircraft models. Each aircraft is annotated with tight bounding boxes and hierarchical model labels. Model labels are organized into four levels: model, variant, family, and manufacturer. The dataset is split equally into training, validation, and test subsets. It was created by multiple researchers and photographers, and the images are for non‑commercial research use only.
