M2VDet: Midpoints-to-Vertices Detection of Oriented Objects in Remote-Sensing Images

Xueru Xu, Zhong Chen, Guoyou Wang, He Deng, Longji Yu, Qimeng Chen

Abstract: Oriented object detectors provide many scientific solutions for object detection tasks in remote-sensing scenes. Among them, anchor-free oriented object detectors have attracted much attention because of their flexibility and conciseness in recent years. However, the positive sampling strategy adopted by most modern anchor-free oriented detectors is inadequate to reflect remote-sensing objects characterized by arbitrary orientation, densely packed, and extensive scale variation. Besides, there is still a lack of a general representation to describe oriented objects shaped with arbitrary convex quadrilaterals. In this article, we present a one-stage anchor-free oriented object detector, called M2VDet, which focuses on designing the sampling strategy and oriented bounding box (OBB) description. First, an adaptive sampling strategy is proposed to generate positive sample spaces, which take full account of objects’ characteristics like scale, shape, and orientation. The resulting candidate regions provide a superior space constraint to construct the heatmap, and then 2-D oriented Gaussian distribution is employed to generate prior labels for positive candidates. Second, a general OBB representation is designed by adopting the midpoints-to-vertices strategy, which provides a unified approach to describe the arbitrary convex quadrilaterals without brutal approximation. Extensive experimental results on four public datasets (e.g., DOTA, HRSC2016, UCAS-AOD, and SODA-A) demonstrate that M2VDet has a simple pipeline, but it can achieve more robust and available performance when compared with state-of-the-art baseline detectors, especially on the detection of densely packed objects with arbitrary orientation as well as small objects.