Consistency Regularization for Deep Face Anti-Spoofing
نویسندگان
چکیده
Face anti-spoofing (FAS) plays a crucial role in securing face recognition systems. Empirically, given an image, model with more consistent output on different views (i.e., augmentations) of this image usually performs better. Motivated by exciting observation, we conjecture that encouraging feature consistency may be promising way to boost FAS models. In paper, explore thoroughly enhancing both Embedding-level and Prediction-level Consistency Regularization (EPCR) FAS. Specifically, at the embedding level, design dense similarity loss maximize similarities between all positions two intermediate maps self-supervised fashion; while prediction optimize mean square error predictions views. Notably, our EPCR is free annotations can directly integrate into semi-supervised learning schemes. Considering application scenarios, further five diverse protocols measure techniques. We conduct extensive experiments show significantly improve performance several supervised tasks benchmark datasets. The codes are available https://github.com/clks-wzz/EPCR .
منابع مشابه
Learning Deep Models for Face Anti-Spoofing: Binary or Auxiliary Supervision
Face anti-spoofing is crucial to prevent face recognition systems from a security breach. Previous deep learning approaches formulate face anti-spoofing as a binary classification problem. Many of them struggle to grasp adequate spoofing cues and generalize poorly. In this paper, we argue the importance of auxiliary supervision to guide the learning toward discriminative and generalizable cues....
متن کاملDeep Representations for Iris, Face, and Fingerprint Spoofing Detection
In biometrics, Fingerprint is widely used in identification of individual’s identity. Biometric recognition is leading technology for identification and security systems. Fingerprint has unique identification among all other biometric modalities. Use of the fingerprints as biometric characteristics is extensively used and developed for fingerprint recognition in forensic, civilian and commercia...
متن کاملLearn Convolutional Neural Network for Face Anti-Spoofing
Though having achieved some progresses, the hand-crafted texture features, e.g., LBP [23], LBP-TOP [11] are still unable to capture the most discriminative cues between genuine and fake faces. In this paper, instead of designing feature by ourselves, we rely on the deep convolutional neural network (CNN) to learn features of high discriminative ability in a supervised manner. Combined with some...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: IEEE Transactions on Information Forensics and Security
سال: 2023
ISSN: ['1556-6013', '1556-6021']
DOI: https://doi.org/10.1109/tifs.2023.3235581