Face Anti Spoofing
Face anti-spoofing (FAS) aims to secure facial recognition systems by detecting fake faces (spoofs) presented as legitimate identities. Current research heavily focuses on improving the generalization of FAS models across diverse domains (different cameras, lighting, attack types), employing architectures like Vision Transformers (ViTs) and Convolutional Neural Networks (CNNs), often enhanced by techniques such as data augmentation, multimodal fusion, and self-supervised learning. The development of robust and generalized FAS methods is crucial for the widespread adoption of secure facial recognition technologies in various applications, from access control to financial transactions.
Papers
Learning Polysemantic Spoof Trace: A Multi-Modal Disentanglement Network for Face Anti-spoofing
Kaicheng Li, Hongyu Yang, Binghui Chen, Pengyu Li, Biao Wang, Di Huang
Cyclically Disentangled Feature Translation for Face Anti-spoofing
Haixiao Yue, Keyao Wang, Guosheng Zhang, Haocheng Feng, Junyu Han, Errui Ding, Jingdong Wang