VerifIoU -- Robustness of Object Detection to Perturbations

TL;DR

This paper presents IBP IoU, a new formal verification method for object detection models that improves robustness against perturbations, demonstrated on runway detection and digit recognition tasks.

Contribution

Introduces IBP IoU, a novel Interval Bound Propagation approach for verifying object detection models' robustness with respect to the IoU metric.

Findings

IBP IoU outperforms baseline in accuracy and stability

Open source implementation available for verification tasks

Effective on runway detection and digit recognition cases

Abstract

We introduce a novel Interval Bound Propagation (IBP) approach for the formal verification of object detection models, specifically targeting the Intersection over Union (IoU) metric. The approach has been implemented in an open source code, named IBP IoU, compatible with popular abstract interpretation based verification tools. The resulting verifier is evaluated on landing approach runway detection and handwritten digit recognition case studies. Comparisons against a baseline (Vanilla IBP IoU) highlight the superior performance of IBP IoU in ensuring accuracy and stability, contributing to more secure and robust machine learning applications.