Arbitrary-oriented object detection (AOOD) has been widely applied to locate and classify objects with diverse orientations in remote sensing images. However, the inconsistent features for the localization and classification tasks in AOOD models may lead to ambiguity and low-quality object predictions, which constrains the detection performance. In this paper, an AOOD method called task-wise sampling convolutions (TS-Conv) is proposed. TS-Conv adaptively samples task-wise features from respective sensitive regions and maps these features together in alignment to guide a dynamic label assignment for better predictions. Specifically, sampling positions of the localization convolution in TS-Conv is supervised by the oriented bounding box (OBB) prediction associated with spatial coordinates. While sampling positions and convolutional kernel of the classification convolution are designed to be adaptively adjusted according to different orientations for improving the orientation robustness of features. Furthermore, a dynamic task-aware label assignment (DTLA) strategy is developed to select optimal candidate positions and assign labels dynamicly according to ranked task-aware scores obtained from TS-Conv. Extensive experiments on several public datasets covering multiple scenes, multimodal images, and multiple categories of objects demonstrate the effectiveness, scalability and superior performance of the proposed TS-Conv.

翻译：任意方向目标检测（AOOD）已广泛应用于遥感图像中不同朝向目标的定位与分类任务。然而，AOOD模型中定位与分类任务特征的不一致性可能导致目标预测模糊且质量低下，从而制约检测性能。本文提出一种称为逐任务采样卷积（TS-Conv）的AOOD方法。TS-Conv从各自敏感区域自适应采样任务特异性特征，并将这些特征对齐映射，以引导动态标签分配生成更优预测结果。具体而言，TS-Conv中定位卷积的采样位置由与空间坐标关联的有向边界框（OBB）预测进行监督；而分类卷积的采样位置与卷积核被设计为根据不同朝向自适应调整，以提升特征的朝向鲁棒性。进一步地，本文提出动态任务感知标签分配（DTLA）策略，通过根据TS-Conv生成的按序排列的任务感知分数，选择最优候选位置并动态分配标签。在涵盖多场景、多模态图像及多类别目标的多个公开数据集上的大量实验表明，所提出的TS-Conv方法具有有效性、可扩展性与优越性能。