VerifIoU -- Robustness of Object Detection to Perturbations
This work addresses the need for more secure and robust machine learning applications, particularly in safety-critical domains like aviation, though it appears incremental as an extension of existing IBP methods.
The paper tackles the problem of formally verifying object detection models against perturbations using a novel Interval Bound Propagation approach for the Intersection over Union metric, achieving superior performance in accuracy and stability compared to a baseline in case studies like landing approach runway detection and handwritten digit recognition.
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.