Evaluating Robustness of Deep Image Super-Resolution against Adversarial Attacks
This work addresses a security problem for computer vision applications relying on super-resolution, but it is incremental as it focuses on evaluating existing methods rather than proposing new defenses.
The paper investigated the vulnerability of deep learning-based super-resolution methods to adversarial attacks, finding that state-of-the-art methods are highly susceptible, with attacks causing significant deterioration in super-resolved images without noticeable distortion in low-resolution inputs.
Single-image super-resolution aims to generate a high-resolution version of a low-resolution image, which serves as an essential component in many computer vision applications. This paper investigates the robustness of deep learning-based super-resolution methods against adversarial attacks, which can significantly deteriorate the super-resolved images without noticeable distortion in the attacked low-resolution images. It is demonstrated that state-of-the-art deep super-resolution methods are highly vulnerable to adversarial attacks. Different levels of robustness of different methods are analyzed theoretically and experimentally. We also present analysis on transferability of attacks, and feasibility of targeted attacks and universal attacks.