Detecting Rotated Objects as Gaussian Distributions and Its 3-D Generalization

TL;DR

This paper introduces a novel Gaussian distribution-based approach for rotated object detection, replacing traditional bounding box methods, leading to improved high-precision detection and extending to 3-D with broad applicability.

Contribution

The paper proposes modeling rotated objects as Gaussian distributions and a new regression loss based on Kullback-Leibler Divergence, addressing fundamental limitations of traditional bounding box methods.

Findings

Outperforms existing methods on twelve datasets

Effectively handles boundary discontinuity and square-like issues

Extends to 3-D object detection with tailored algorithms

Abstract

Existing detection methods commonly use a parameterized bounding box (BBox) to model and detect (horizontal) objects and an additional rotation angle parameter is used for rotated objects. We argue that such a mechanism has fundamental limitations in building an effective regression loss for rotation detection, especially for high-precision detection with high IoU (e.g. 0.75). Instead, we propose to model the rotated objects as Gaussian distributions. A direct advantage is that our new regression loss regarding the distance between two Gaussians e.g. Kullback-Leibler Divergence (KLD), can well align the actual detection performance metric, which is not well addressed in existing methods. Moreover, the two bottlenecks i.e. boundary discontinuity and square-like problem also disappear. We also propose an efficient Gaussian metric-based label assignment strategy to further boost the performance. Interestingly, by analyzing the BBox parameters' gradients under our Gaussian-based KLD loss, we show that these parameters are dynamically updated with interpretable physical meaning, which help explain the effectiveness of our approach, especially for high-precision detection. We extend our approach from 2-D to 3-D with a tailored algorithm design to handle the heading estimation, and experimental results on twelve public datasets (2-D/3-D, aerial/text/face images) with various base detectors show its superiority.