Fast and precise object detection for hgigh-resolution aerial images has been a challenging task over the years. Due to the sharp variations in object scale, rotation, and aspect ratio, most existing methods are inefficient and imprecise. In this paper, we propose a different approach polar method. We locate an object by centrepoint, direct it by four polar angles, and measure it by polar ratio system. Our polar coordinate-based method, PolarDet, is a faster, simpler, and more accurate one-stage object detector. Also, our detector introduces a sub-pixel centre semantic structure to further improve classifying veracity. PolarDet achieves nearly all state-ofthe-art (SOTA) performance in aerial object detection tasks with faster inference speed. In detail, our approach obtains the SOTA results on authoritative remote sensing object detection datasets DOTA, UCAS-AOD, and HRSC2016 with 76.64% mAP (mean average precision), 97.01% mAP, and 90.46% mAP respectively. Most noticeably, our PolarDet gets the best performance and reaches the fastest speed (32fps) at the UCAS-AOD dataset.

No abstract

No abstract

This report summarizes the results of Learning to Understand Aerial Images (LUAI) 2021 challenge held on ICCV'2021, which focuses on object detection and semantic segmentation in aerial images. Using and datasets, this challenge proposes three tasks for oriented object detection, horizontal object detection, and semantic segmentation of common categories in aerial images. This challenge received a total of 146 registrations on the three tasks. Through the challenge, we hope to draw attention from a wide range of communities and call for more efforts on the problems of learning to understand aerial images.