Adversarial Attack and Defense for Dehazing Networks

The research on single image dehazing task has been widely explored. However, as far as we know, no comprehensive study has been conducted on the robustness of the well-trained dehazing models. Therefore, there is no evidence that the dehazing networks can resist malicious attacks. In this paper, we focus on designing a group of attack methods based on first order gradient to verify the robustness of the existing dehazing algorithms. By analyzing the general goal of image dehazing task, five attack methods are proposed, which are prediction, noise, mask, ground-truth and input attack. The corresponding experiments are conducted on six datasets with different scales. Further, the defense strategy based on adversarial training is adopted for reducing the negative effects caused by malicious attacks. In summary, this paper defines a new challenging problem for image dehazing area, which can be called as adversarial attack on dehazing networks (AADN). Code is available at https://github.com/guijiejie/AADN.