Suppressing Spoof-Irrelevant Factors for Domain-Agnostic Face Anti-Spoofing

Abstract

Face anti-spoofing aims to prevent false authentications of face recognition systems by distinguishing whether an image is originated from a human face or a spoof medium. In this work, we note that images from unseen domains having different spoof-irrelevant factors (e.g., background patterns and subject) induce domain shift between source and target distributions. Also, when the same SiFs are shared by the spoof and genuine images, they show a higher level of visual similarity and this hinders accurate face anti-spoofing. Hence, we aim to minimize the discrepancies among different domains via alleviating the effects of SiFs, and achieve improvements in generalization to unseen domains. To realize our goal, we propose a novel method called a Doubly Adversarial Suppression Network (DASN) that is trained to neglect the irrelevant factors and to focus more on faithful task-relevant factors. Our DASN consists of two types of adversarial learning schemes. In the first adversarial learning scheme, multiple SiFs are suppressed by deploying multiple discrimination heads that are trained against an encoder. In the second adversarial learning scheme, each of the discrimination heads is also adversarially trained to suppress a spoof factor, and the group of the secondary spoof classifier and the encoder aims to intensify the spoof factor by overcoming the suppression. We evaluate the proposed method on four public benchmark datasets, and achieve remarkable evaluation results in generalizing to unseen domains. The results demonstrate the effectiveness of the proposed method.

Adaptive Transformers for Robust Few-shot Cross-domain Face Anti-spoofing

“…Domain generalization for face anti-spoofing aims to learn a model from multiple source datasets, and the model should generalize to the unseen target dataset. Several approaches [53,22,52,24] based on adversarial training and triplet loss have been developed to learn a shared feature space for multiple source domains that can generalize to the target domain. On the other hand, meta-learning formulations [54,7,62] are exploited to simulate the domain shift at training time to learn a representative feature space.…”

“…Explanation about the benchmark. We follow the typical cross-domain evaluation setting that is widely used in face anti-spoofing literature [53,22,52,24,54,7,62,61,38,73,37,35]. We note that the zero-shot benchmark proposed in [40] has been previously retrieved and is no longer available.…”

Deep Learning meets Liveness Detection: Recent Advancements and Challenges

“…Although most domain generalization methods exploit datasets' source to find a generalizable feature space, they do not make use of additional information that comes with most FAS dataset. However, Kim et al [110] exploit such information by forcing the feature extractor to capture faithful features of FAS and discarding irrelevant features, such as, identity, acquisition device, environment, etc. The training strategy is comprised of two steps: learning discriminative features while discarding spoof-irrelevant factors (SiFs), improving the encoder discriminability.…”