Category-Aware Dynamic Label Assignment with High-Quality Oriented Proposal

TL;DR

This paper introduces a novel category-aware dynamic label assignment method with a high-quality oriented proposal generator, improving aerial object detection accuracy by addressing angle discontinuities and proposal label issues.

Contribution

It proposes a complex plane-based OBB representation, a conformer RPN head for angle prediction, and a dynamic label assignment strategy based on category feedback, enhancing detection performance.

Findings

Achieved state-of-the-art mAP scores on multiple aerial datasets.

Reduced parameter tuning and computational costs.

Enhanced proposal quality and label assignment accuracy.

Abstract

Objects in aerial images are typically embedded in complex backgrounds and exhibit arbitrary orientations. When employing oriented bounding boxes (OBB) to represent arbitrary oriented objects, the periodicity of angles could lead to discontinuities in label regression values at the boundaries, inducing abrupt fluctuations in the loss function. To address this problem, an OBB representation based on the complex plane is introduced in the oriented detection framework, and a trigonometric loss function is proposed. Moreover, leveraging prior knowledge of complex background environments and significant differences in large objects in aerial images, a conformer RPN head is constructed to predict angle information. The proposed loss function and conformer RPN head jointly generate high-quality oriented proposals. A category-aware dynamic label assignment based on predicted category feedback is proposed to address the limitations of solely relying on IoU for proposal label assignment. This method makes negative sample selection more representative, ensuring consistency between classification and regression features. Experiments were conducted on four realistic oriented detection datasets, and the results demonstrate superior performance in oriented object detection with minimal parameter tuning and time costs. Specifically, mean average precision (mAP) scores of 82.02%, 71.99%, 69.87%, and 98.77% were achieved on the DOTA-v1.0, DOTA-v1.5, DIOR-R, and HRSC2016 datasets, respectively.