Consistent Semantic Attacks on Optical Flow

Tom Koren, Lior Talker, Michael Dinerstein, Ran Vitek; Proceedings of the Asian Conference on Computer Vision (ACCV), 2022, pp. 1658-1674


We present a novel approach for semantically targeted adversarial attacks on Optical Flow. In such attacks the goal is to corrupt the flow predictions of a specific object category or instance. Usually, an attacker seeks to hide the adversarial perturbations in the input. However, a quick scan of the output reveals the attack. In contrast, our method helps to hide the attacker's intent in the output flow as well. We achieve this thanks to a regularization term that encourages off-target consistency. We perform extensive tests on leading optical flow models to demonstrate the benefits of our approach in both white-box and blackbox settings. Also, we demonstrate the effectiveness of our attack on subsequent tasks that depend on the optical flow.

Related Material

[pdf] [supp] [arXiv]
@InProceedings{Koren_2022_ACCV, author = {Koren, Tom and Talker, Lior and Dinerstein, Michael and Vitek, Ran}, title = {Consistent Semantic Attacks on Optical Flow}, booktitle = {Proceedings of the Asian Conference on Computer Vision (ACCV)}, month = {December}, year = {2022}, pages = {1658-1674} }