IoUCert: Robustness Verification for Anchor-based Object Detectors

2026-03-03

AI Summary

This work addresses the challenge of formally verifying the robustness of anchor-based object detectors, which has been hindered by the nonlinearity of coordinate transformations and the complexity of the Intersection-over-Union (IoU) metric. Focusing on single-object localization, the authors propose a novel coordinate transformation that avoids accuracy-degrading relaxations of bounding box prediction functions and instead directly optimizes IoU bounds based on anchor offsets. By designing a differentiable and tight interval bound propagation (IBP) mechanism, they achieve, for the first time, formal robustness verification under input perturbations for mainstream anchor-based detectors such as SSD, YOLOv2, and YOLOv3. This breakthrough overcomes the longstanding barrier that complex detection architectures could not be rigorously verified using existing formal methods.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Adversarial Learning & Robustness
• Intelligent Robots: State Estimation

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Web Mining and Content Analysis: Web data provenance, reliability, and authenticity
• Security and Privacy: Data transparency and provenance

Abstract

While formal robustness verification has seen significant success in image classification, scaling these guarantees to object detection remains notoriously difficult due to complex non-linear coordinate transformations and Intersection-over-Union (IoU) metrics. We introduce {\sc \sf IoUCert}, a novel formal verification framework designed specifically to overcome these bottlenecks in foundational anchor-based object detection architectures. Focusing on the object localisation component in single-object settings, we propose a coordinate transformation that enables our algorithm to circumvent precision-degrading relaxations of non-linear box prediction functions. This allows us to optimise bounds directly with respect to the anchor box offsets which enables a novel Interval Bound Propagation method that derives optimal IoU bounds. We demonstrate that our method enables, for the first time, the robustness verification of realistic, anchor-based models including SSD, YOLOv2, and YOLOv3 variants against various input perturbations.

Problem

• robustness verification
• object detection
• anchor-based
• Intersection-over-Union
• formal verification

Innovation

• formal verification
• object detection
• Intersection-over-Union (IoU)
• Interval Bound Propagation
• anchor-based models