Unveiling hidden vulnerabilities in digital human generation via adversarial attacks

Zhiying Li, Yeying Jin, Fan Shen, Zhi Liu, Weibin Chen, Pengju Zhang, Xiaomei Zhang, Boyu Chen, Michael Shen, Kejian Wu, Zhaoxin Fan, Jin Dong

Abstract: Highlights•Proposing Tangible Attack (TBA) on the expressive human pose and shape estimation models for digital human security.•Proposing Dual Heterogeneous Noise Generator using Variational Autoencoder and Diffusion for adversaries.•Proposing a novel adversarial loss to optimally enhance its attack performance.•Experiments show our attack is effective on all expressive human pose and shape estimation models.