PointOBB-v3: Expanding Performance Boundaries of Single Point-Supervised Oriented Object Detection

TL;DR

PointOBB-v3 advances single point-supervised oriented object detection by integrating multi-view strategies, scale and angle modules, and an end-to-end approach, achieving significant accuracy improvements across multiple datasets.

Contribution

It introduces a novel end-to-end framework with scale and angle modules, eliminating pseudo-labels and enhancing detection accuracy in point-supervised OOD.

Findings

Achieves an average accuracy improvement of 3.56% over previous methods.

Effectively estimates object scale using SSC and SSFF modules.

Provides a robust angle prediction via symmetry-based self-supervised learning.

Abstract

With the growing demand for oriented object detection (OOD), recent studies on point-supervised OOD have attracted significant interest. In this paper, we propose PointOBB-v3, a stronger single point-supervised OOD framework. Compared to existing methods, it generates pseudo rotated boxes without additional priors and incorporates support for the end-to-end paradigm. PointOBB-v3 functions by integrating three unique image views: the original view, a resized view, and a rotated/flipped (rot/flp) view. Based on the views, a scale augmentation module and an angle acquisition module are constructed. In the first module, a Scale-Sensitive Consistency (SSC) loss and a Scale-Sensitive Feature Fusion (SSFF) module are introduced to improve the model's ability to estimate object scale. To achieve precise angle predictions, the second module employs symmetry-based self-supervised learning. Additionally, we introduce an end-to-end version that eliminates the pseudo-label generation process by integrating a detector branch and introduces an Instance-Aware Weighting (IAW) strategy to focus on high-quality predictions. We conducted extensive experiments on the DIOR-R, DOTA-v1.0/v1.5/v2.0, FAIR1M, STAR, and RSAR datasets. Across all these datasets, our method achieves an average improvement in accuracy of 3.56% in comparison to previous state-of-the-art methods. The code will be available at https://github.com/ZpyWHU/PointOBB-v3.