Deep Spatial Gradient and Temporal Depth Learning for Face Anti-Spoofing

Abstract

Face anti-spoofing is critical to the security of face recognition systems. Depth supervised learning has been proven as one of the most effective methods for face antispoofing. Despite the great success, most previous works still formulate the problem as a single-frame multi-task one by simply augmenting the loss with depth, while neglecting the detailed fine-grained information and the interplay between facial depths and moving patterns. In contrast, we design a new approach to detect presentation attacks from multiple frames based on two insights: 1) detailed discriminative clues (e.g., spatial gradient magnitude) between living and spoofing face may be discarded through stacked vanilla convolutions, and 2) the dynamics of 3D moving faces provide important clues in detecting the spoofing faces. The proposed method is able to capture discriminative details via Residual Spatial Gradient Block (RSGB) and encode spatio-temporal information from Spatio-Temporal Propagation Module (STPM) efficiently. Moreover, a novel Contrastive Depth Loss is presented for more accurate depth supervision. To assess the efficacy of our method, we also collect a Double-modal Anti-spoofing Dataset (DMAD) which provides actual depth for each sample. The experiments demonstrate that the proposed approach achieves state-of-the-art results on five benchmark datasets including OULU-NPU, SiW, CASIA-MFSD, Replay-Attack, and the new DMAD. Codes will be available at https://github.com/clks-wzz/ FAS-SGTD.

NAS-FAS: Static-Dynamic Central Difference Network Search for Face Anti-Spoofing

“…All recorded subjects are Chinese people (21 males and 27 females). In terms of the age distribution, most of the subjects are within the range [20,30) and [50,60) years old. The youngest and eldest age is 23 and 62, respectively.…”

Multi-Modal Face Anti-Spoofing Based on Central Difference Networks

“…In contrast, auxiliary depth supervised FAS methods [2,22] are introduced to learn more detailed information effectively. On the other hand, several video level CNN methods are presented to exploit the dynamic spatiotemporal [33,34,19] or rPPG [17,22,18,36,37,31] features for PAD. Despite achieving state-of-the-art performance, single-modal methods are easily influenced by unseen domain shift (e.g., cross ethnicity and cross attack types) and not robust for challenging cases (e.g., harsh environment and realistic attacks).…”

“…where H, W denote the height and width of the binary mask, respectively, and B pre and B gt mean the predicted grayscale mask and ground truth binary mask, respectively. Moreover, for the sake of fine-grained supervision needs in FAS task, contrastive depth loss (CDL) L CDL [33] is considered to help the networks learn more detailed features. CDL can be formulated as…”

“…In recent years, several hand-crafted feature based [3,4,7,15,28,27] and deep learning based [38,33,29,22,12,34,2,8,9] methods have been proposed for presentation attack detection (PAD). On one hand, the classical hand- * denotes corresponding author [21].…”

Efficient Face Spoofing Detection With Flash

“…Under a simple live/spoof binary classification settings, resulting two representations separated by a classification boundary can have mixed information, such as subject ID. Several neural networks are proposed to disentangle live/spoof class signature from the unwanted information [48], [49], [50] Because most of the presentation attacks are conducted under visible-light domain, using multimodal input or non-visible modality (e.g., depth, rPPG, SWIR) are useful for increasing PAD performance [6], [51], [52], [53], [54], [55], [56], [57], [58], [59], [60], [61], [62].…”