Temporal sparse adversarial attack on sequence-based gait recognition

Gait recognition is widely used in social security applications due to its advantages long-distance human identification. Recently, sequence-based methods have achieved high accuracy by learning abundant temporal and spatial information. However, their robustness under adversarial attacks an open world has not been clearly explored. In this paper, we demonstrate that the state-of-the-art gait model vulnerable such attacks. To end, propose a novel sparse attack method. Different from previous additive noise models which add perturbations on original samples, employ generative network based architecture semantically generate high-quality silhouettes or video frames. Moreover, sparsely substituting inserting few silhouettes, proposed method ensures imperceptibility achieves strong ability. The experimental results show if only one-fortieth of frames are attacked, target drops dramatically.