Black-box Adversarial Attacks on Monocular Depth Estimation Using Evolutionary Multi-objective Optimization
This research addresses the vulnerability of monocular depth estimation DNNs to black-box adversarial attacks, which is a significant concern for the robustness and security of these systems in real-world applications.
This paper introduces a black-box adversarial attack method for monocular depth estimation DNNs, operating with only output depth maps. The method successfully attacked two different DNNs trained on indoor and outdoor scenes without requiring knowledge of their architecture or training data.
This paper proposes an adversarial attack method to deep neural networks (DNNs) for monocular depth estimation, i.e., estimating the depth from a single image. Single image depth estimation has improved drastically in recent years due to the development of DNNs. However, vulnerabilities of DNNs for image classification have been revealed by adversarial attacks, and DNNs for monocular depth estimation could contain similar vulnerabilities. Therefore, research on vulnerabilities of DNNs for monocular depth estimation has spread rapidly, but many of them assume white-box conditions where inside information of DNNs is available, or are transferability-based black-box attacks that require a substitute DNN model and a training dataset. Utilizing Evolutionary Multi-objective Optimization, the proposed method in this paper analyzes DNNs under the black-box condition where only output depth maps are available. In addition, the proposed method does not require a substitute DNN that has a similar architecture to the target DNN nor any knowledge about training data used to train the target model. Experimental results showed that the proposed method succeeded in attacking two DNN-based methods that were trained with indoor and outdoor scenes respectively.