Adversarial Attack on Video Retrieval
Ying Zou, Chenglong Zhao, Bingbing Ni · 2020
Recently adversarial examples have been reported to reveal the fragility of deep learning models. However, most adversarial attacks focus on classification task and less attention has been paid to retrieval task. In this paper, we are the first to investigate adversarial examples on the video retrieval system in both non-targeted and targeted attack terms for copyright protection. Specifically, a triplet scheme is developed to take query-relevant and query-target pair-wise relationships together to enhance the attack performance. We evaluated the proposed method on the most commonly used video retrieval dataset CC_WEB_VIDEO, and successfully attack three popular video retrieval systems.