Temporal sparse adversarial attack on sequence-based gait recognition

作者：

Highlights：

• We propose a novel temporal sparse adversarial attack method on gait recognition models.

• The proposed method achieves good imperceptibility and a high attack success rate.

• Our method is capable of attacking on biometric samples to make a practical threat.

• We identify the adversarial vulnerability of the state-of-the-art sequence-based gait recognition model.

摘要

•We propose a novel temporal sparse adversarial attack method on gait recognition models.•The proposed method achieves good imperceptibility and a high attack success rate.•Our method is capable of attacking on biometric samples to make a practical threat.•We identify the adversarial vulnerability of the state-of-the-art sequence-based gait recognition model.

论文关键词：Adversarial attack,Gait recognition,Temporal sparsity

论文评审过程：Received 2 February 2022, Revised 21 June 2022, Accepted 4 September 2022, Available online 7 September 2022, Version of Record 23 September 2022.

论文官网地址：https://doi.org/10.1016/j.patcog.2022.109028