Projection-based Adversarial Attack using Physics-in-the-Loop Optimization for Monocular Depth Estimation

TL;DR

This paper introduces a novel projection-based adversarial attack method for monocular depth estimation models, using physics-in-the-loop optimization to create realistic perturbations that cause depth estimation failures.

Contribution

It presents a new attack technique combining projection and physics-in-the-loop optimization to effectively generate adversarial examples for monocular depth estimation models.

Findings

Successfully causes depth misestimations in models

Objects partially disappear in attacked scenes

Demonstrates vulnerability of DNN-based MDE models

Abstract

Deep neural networks (DNNs) remain vulnerable to adversarial attacks that cause misclassification when specific perturbations are added to input images. This vulnerability also threatens the reliability of DNN-based monocular depth estimation (MDE) models, making robustness enhancement a critical need in practical applications. To validate the vulnerability of DNN-based MDE models, this study proposes a projection-based adversarial attack method that projects perturbation light onto a target object. The proposed method employs physics-in-the-loop (PITL) optimization -- evaluating candidate solutions in actual environments to account for device specifications and disturbances -- and utilizes a distributed covariance matrix adaptation evolution strategy. Experiments confirmed that the proposed method successfully created adversarial examples that lead to depth misestimations, resulting in parts of objects disappearing from the target scene.