The paper introduces a large-scale dataset for object detection in aerial images, named DOTA (Dataset for Object deTection in Aerial images). DOTA aims to address the challenges in object detection in Earth Vision, such as the variation in scale, orientation, and shape of objects, as well as the scarcity of well-annotated datasets. The dataset consists of 2806 aerial images, each containing objects of various scales, orientations, and shapes, annotated by experts using 15 common object categories. Each image is approximately 4000 × 4000 pixels in size, and the dataset contains 188,282 instances, each labeled with an arbitrary quadrilateral bounding box. The authors evaluate state-of-the-art object detection algorithms on DOTA, demonstrating its representativeness and challenge for real Earth Vision applications. DOTA is the largest annotated object dataset with a wide variety of categories in Earth Vision and can be used to develop and evaluate object detectors in aerial images. The dataset is available for public use and will be updated to reflect evolving real-world conditions.The paper introduces a large-scale dataset for object detection in aerial images, named DOTA (Dataset for Object deTection in Aerial images). DOTA aims to address the challenges in object detection in Earth Vision, such as the variation in scale, orientation, and shape of objects, as well as the scarcity of well-annotated datasets. The dataset consists of 2806 aerial images, each containing objects of various scales, orientations, and shapes, annotated by experts using 15 common object categories. Each image is approximately 4000 × 4000 pixels in size, and the dataset contains 188,282 instances, each labeled with an arbitrary quadrilateral bounding box. The authors evaluate state-of-the-art object detection algorithms on DOTA, demonstrating its representativeness and challenge for real Earth Vision applications. DOTA is the largest annotated object dataset with a wide variety of categories in Earth Vision and can be used to develop and evaluate object detectors in aerial images. The dataset is available for public use and will be updated to reflect evolving real-world conditions.