R2Det: Exploring Relaxed Rotation Equivariance in 2D object detection

2024-08-21

AI Summary

Traditional group-equivariant convolutions (GConvs) for 2D object detection rely strictly on cyclic groups $C_n$, limiting their ability to handle broken rotational symmetry and non-rigid deformations. To address this, we propose the relaxed rotation-equivariant group $R_n$ and its associated $R^2$GConv operator, which introduces only $4n$ additional parameters while relaxing the rigid constraints of $C_n$, preserving equivariance and enhancing modeling capacity for realistic asymmetric scenes. Based on this, we design R2Det—a two-stage, rotation-robust detector featuring a customized R2Net backbone. Experiments validate the efficacy of $R^2$GConv on image classification; R2Det achieves state-of-the-art performance across multiple 2D detection benchmarks, with significant improvements in robustness to geometric perturbations and cross-domain generalization.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Adversarial Learning & Robustness
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Group Equivariant Convolution (GConv) empowers models to explore underlying symmetry in data, improving performance. However, real-world scenarios often deviate from ideal symmetric systems caused by physical permutation, characterized by non-trivial actions of a symmetry group, resulting in asymmetries that affect the outputs, a phenomenon known as Symmetry Breaking. Traditional GConv-based methods are constrained by rigid operational rules within group space, assuming data remains strictly symmetry after limited group transformations. This limitation makes it difficult to adapt to Symmetry-Breaking and non-rigid transformations. Motivated by this, we mainly focus on a common scenario: Rotational Symmetry-Breaking. By relaxing strict group transformations within Strict Rotation-Equivariant group $mathbf{C}_n$, we redefine a Relaxed Rotation-Equivariant group $mathbf{R}_n$ and introduce a novel Relaxed Rotation-Equivariant GConv (R2GConv) with only a minimal increase of $4n$ parameters compared to GConv. Based on R2GConv, we propose a Relaxed Rotation-Equivariant Network (R2Net) as the backbone and develop a Relaxed Rotation-Equivariant Object Detector (R2Det) for 2D object detection. Experimental results demonstrate the effectiveness of the proposed R2GConv in natural image classification, and R2Det achieves excellent performance in 2D object detection with improved generalization capabilities and robustness. The code is available in exttt{https://github.com/wuer5/r2det}.

Problem

• Addressing Symmetry Breaking in 2D object detection
• Relaxing rigid group transformations for better adaptation
• Improving generalization and robustness in object detection

Innovation

• Relaxed Rotation-Equivariant GConv introduced
• Minimal parameter increase in R2GConv
• R2Det enhances 2D object detection