Object detection in aerial images using DOTA dataset: A survey

• Published in: International journal of applied earth observation and geoinformation Vol. 134; p. 104208
• Main Authors: Chen, Ziyi, Wang, Huayou, Wu, Xinyuan, Wang, Jing, Lin, Xinrui, Wang, Cheng, Gao, Kyle, Chapman, Michael, Li, Dilong
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.11.2024; Elsevier
• Subjects: data collection; Deep learning; DOTA dataset; Object detection; RSIs; spatial data; surveys; Deep learning; RSIs; Object detection; DOTA dataset
• ISSN: 1569-8432
• DOI: 10.1016/j.jag.2024.104208

•Presents a comprehensive overview of literature studies related to the DOTA dataset for the first time.•Revisits other prevalent datasets related to RSIs, and offers a detailed comparative analysis against the DOTA dataset.•Elaborates on both traditional object detection techniques and those rooted in deep learning specific to remote sensing imagery. In recent years, the Dataset for Object deTection in Aerial images (DOTA) dataset has played a pivotal role in advancing object detection in aerial images (ODAI). Despite its significance, there hasn’t been a comprehensive review summarizing its research developments. Addressing this gap, this paper offers the first comprehensive overview on the subject. Within this review, we begin by examining prevalent object detection datasets of natural scene images alongside object detection datasets of remote sensing images (RSIs). We then present an in-depth comparative analysis between these datasets and the DOTA dataset, supported by numerous charts and tables. We proceed to outline both traditional techniques for ODAI and methods rooted in deep learning. Subsequently, we provide a recap of the latest advancements in the field achieved using the DOTA dataset. Concluding our review, we delve into the current challenges facing ODAI and propose potential future research directions.