STA: Adversarial Attacks on Siamese Trackers
This work addresses security risks in visual tracking systems, particularly for applications like surveillance and autonomous driving, by introducing a novel attack method, though it is incremental in the broader context of adversarial machine learning.
The paper tackles the vulnerability of Siamese trackers to adversarial attacks by analyzing their weaknesses and proposing an end-to-end pipeline to generate 3D adversarial examples, which successfully lower tracking accuracy and cause tracker drift.
Recently, the majority of visual trackers adopt Convolutional Neural Network (CNN) as their backbone to achieve high tracking accuracy. However, less attention has been paid to the potential adversarial threats brought by CNN, including Siamese network. In this paper, we first analyze the existing vulnerabilities in Siamese trackers and propose the requirements for a successful adversarial attack. On this basis, we formulate the adversarial generation problem and propose an end-to-end pipeline to generate a perturbed texture map for the 3D object that causes the trackers to fail. Finally, we conduct thorough experiments to verify the effectiveness of our algorithm. Experiment results show that adversarial examples generated by our algorithm can successfully lower the tracking accuracy of victim trackers and even make them drift off. To the best of our knowledge, this is the first work to generate 3D adversarial examples on visual trackers.