A comprehensive survey of oriented object detection in remote sensing images.

With the rapid development of object detection, it is widely used in many scenes and images. However, the dense arrangement of objects with different dimensions, orientations and aspect ratios in remote sensing and aerial images undoubtedly poses many problems for detection. Anchor-based oriented object detection to maintain rotational invariance has to solve the problem of object orientation dimension and also to consider the calculation of angular periodicity in the regression calculation. To achieve accurate detection of objects, it is necessary to obtain the precise frame surrounding the object and the precise features. Anchor-free methods do not require a predefined anchor, but only need to learn the object feature parameters to get an accurate frame for detection. In this paper we first introduce the technical approaches to object detection, both traditional and deep learning-based methods. Then we summarize the main problems and methods solved in oriented object detection in anchor-based and anchor-free based detection. We present some datasets using oriented bounding box (OBB) annotation that are suitable for oriented object detection, as well as introduce the accepted benchmarks and evaluation metrics for object detection. Finally, we discuss potential trends in oriented object detection for the benefit of researchers who are new to the field.