Misleading attention and classification: An adversarial attack to fool object detection models in the real world

Haotian Zhang, Xu Ma

Abstract: Highlights•We propose an adversarial attack called MACA to fool state-of-the-art detectors.•The MACA realises a universal attack that can transfer among different networks.•We introduce simulated disturbances for the robustness and naturalness of attacking.•We use 3D human models for experiments to demonstrate the effectiveness of MACA.•We accomplish real-world adversarial attacks through trained adversarial patches.