Oriented object detection in remote sensing images (RSIs) is a significant yet challenging Earth Vision task, as the objects in RSIs usually emerge with complicated backgrounds, arbitrary orientations, multi-scale distributions, and dramatic aspect ratio variations. Existing oriented object detectors are mostly inherited from the anchor-based paradigm. However, the prominent performance of high-precision and real-time detection with anchor-based detectors is overshadowed by the design limitations of tediously rotated anchors. By using the simplicity and efficiency of keypoint-based detection, in this work, we extend a keypoint-based detector to the task of oriented object detection in RSIs. Specifically, we first simplify the oriented bounding box (OBB) as a center-based rotated inscribed ellipse (RIE), and then employ six parameters to represent the RIE inside each OBB: the center point position of the RIE, the offsets of the long half axis, the length of the short half axis, and an orientation label. In addition, to resolve the influence of complex backgrounds and large-scale variations, a high-resolution gated aggregation network (HRGANet) is designed to identify the targets of interest from complex backgrounds and fuse multi-scale features by using a gated aggregation model (GAM). Furthermore, by analyzing the influence of eccentricity on orientation error, eccentricity-wise orientation loss (ewoLoss) is proposed to assign the penalties on the orientation loss based on the eccentricity of the RIE, which effectively improves the accuracy of the detection of oriented objects with a large aspect ratio. Extensive experimental results on the DOTA and HRSC2016 datasets demonstrate the effectiveness of the proposed method.

Oriented object detection in optical remote sensing images (ORSIs) is a challenging task since the targets in ORSIs are displayed in an arbitrarily oriented manner and on small scales, and are densely packed. Current state-of-the-art oriented object detection models used in ORSIs primarily evolved from anchor-based and direct regression-based detection paradigms. Nevertheless, they still encounter a design difficulty from handcrafted anchor definitions and learning complexities in direct localization regression. To tackle these issues, in this paper, we proposed a novel multi-sector oriented object detection framework called MS${\mathrm{O}}^2$-Det, which quantizes the scales and orientation prediction of targets in ORSIs via an anchor-free classification-to-regression approach. Specifically, we first represented the arbitrarily oriented bounding box as four scale offsets and angles in four quadrant sectors of the corresponding Cartesian coordinate system. Then, we divided the scales and angle space into multiple discrete sectors and obtained more accurate localization information by a coarse-granularity classification to fine-grained regression strategy. In addition, to decrease the angular-sector classification loss and accelerate the network’s convergence, we designed a smooth angular-sector label (SASL) that smoothly distributes label values with a definite tolerance radius. Finally, we proposed a localization-aided detection score (LADS) to better represent the confidence of a detected box by combining the category-classification score and the sector-selection score. The proposed MS${\mathrm{O}}^2$-Det achieves state-of-the-art results on three widely used benchmarks, including the DOTA, HRSC2016, and UCAS-AOD data sets.

Object detection is regarded as a significant research branch of optical remote sensing image analysis. Aimed at the characteristics of remote sensing image targets such as direction diversity, small scales, and dense arrangements, the state-of-the-art detectors used in optical remote sensing images are primarily extended from some general natural detection approaches based on deep convolutional neural networks (DCNNs). Nevertheless, the detectors intended for remote sensing targets are commonly based on an anchor mechanism and perform regression tasks in the Cartesian coordinate system, which requires complex multiangle anchors, inclined nonmaximum suppression (NMS), and so on. In this letter, a high-resolution polar network (HRPNet) is proposed to detect targets in remote sensing images. As an anchor-free and high-resolution detection network, HRPNet converts the detection of the oriented bounding box into the regression of one polar angle and four polar radii at the corresponding pole points in polar coordinates, which dramatically decreases the computational complexities. As revealed by extensive experiments, the proposed HRPNet shows a competitive advantage on the DOTA and HRRSD data sets.