Aerial scene classification, which aims to automatically label an aerialimage with a specific semantic category, is a fundamental problem forunderstanding high-resolution remote sensing imagery. In recent years, it hasbecome an active task in remote sensing area and numerous algorithms have beenproposed for this task, including many machine learning and data-drivenapproaches. However, the existing datasets for aerial scene classification likeUC-Merced dataset and WHU-RS19 are with relatively small sizes, and the resultson them are already saturated. This largely limits the development of sceneclassification algorithms. This paper describes the Aerial Image Dataset (AID):a large-scale dataset for aerial scene classification. The goal of AID is toadvance the state-of-the-arts in scene classification of remote sensing images.For creating AID, we collect and annotate more than ten thousands aerial sceneimages. In addition, a comprehensive review of the existing aerial sceneclassification techniques as well as recent widely-used deep learning methodsis given. Finally, we provide a performance analysis of typical aerial sceneclassification and deep learning approaches on AID, which can be served as thebaseline results on this benchmark.
展开▼
