A Unified Framework for Iris Anti-Spoofing: Introducing Iris Anti-Spoofing Cross-Domain-Testing Protocol and Masked-MoE Method

TL;DR

This work tackles the critical issue of cross-domain generalization in iris anti-spoofing by introducing the Iris Anti-Spoofing Cross-Domain-Testing (IAS-CDT) Protocol, which benchmarks across races, devices, and multiple datasets. To address cross-domain variability, it proposes Masked-MoE (MMoE), integrating a Mask Mechanism with Cosine Distance Loss to mitigate MoE overfitting and promote inter-expert knowledge sharing, using CLIP as a backbone. Evaluations on 3 sub-protocols show state-of-the-art performance in average, cross-racial, and cross-device settings, highlighting strong generalization capabilities. The framework advances practical iris security by enabling robust anti-spoofing across diverse sensors and populations, with potential for deployment in real-world, device-heterogeneous environments.

Abstract

Iris recognition is widely used in high-security scenarios due to its stability and distinctiveness. However, iris images captured by different devices exhibit certain and device-related consistent differences, which has a greater impact on the classification algorithm for anti-spoofing. The iris of various races would also affect the classification, causing the risk of identity theft. So it is necessary to improve the cross-domain capabilities of the iris anti-spoofing (IAS) methods to enable it more robust in facing different races and devices. However, there is no existing protocol that is comprehensively available. To address this gap, we propose an Iris Anti-Spoofing Cross-Domain-Testing (IAS-CDT) Protocol, which involves 10 datasets, belonging to 7 databases, published by 4 institutions, and collected with 6 different devices. It contains three sub-protocols hierarchically, aimed at evaluating average performance, cross-racial generalization, and cross-device generalization of IAS models. Moreover, to address the cross-device generalization challenge brought by the IAS-CDT Protocol, we employ multiple model parameter sets to learn from the multiple sub-datasets. Specifically, we utilize the Mixture of Experts (MoE) to fit complex data distributions using multiple sub-neural networks. To further enhance the generalization capabilities, we propose a novel method Masked-MoE (MMoE), which randomly masks a portion of tokens for some experts and requires their outputs to be similar to the unmasked experts, which can effectively mitigate the overfitting issue of MoE. For the evaluation, we selected ResNet50, VIT-B/16, CLIP, and FLIP as representative models and benchmarked them under the proposed IAS-CDT Protocol.

Figures (9)

• Figure 1: This figure shows the unified framework for iris anti-spoofing. Including the Iris Anti-Spoofing Cross-Domain-Testing (IAS-CDT) Protocol and Masked-MoE (MMoE) Method.
• Figure 2: This figure shows the development of the iris anti-spoofing (IAS) methods, mainly separated into three phases.
• Figure 3: This figure shows the Train / Test usage of each dataset of the IAS-CDT Protocol. In Protocol-1, all datasets' training and testing sets are involved. In Protocol-2, training sets are H and F, and testing sets involve all 10 subsets. In Protocol-3, for each sub-protocol, we choose one device to test and others to train. And some image samples are shown, including 10 subsets with Real and Fake iris images.
• Figure 4: The Masked Mechanism and Cosine Distance Loss of MMoE. After masking the Dispatch Weights with Mask Matrix, the input tokens calculate with it to mask a part of the information, and output the slots. The slots would be input into the expert, then the cosine distance loss calculates the similarity between the output of each expert.
• Figure 5: This figure shows the combination framework of MMoE with CLIP. The MMoE block is parallel with the MLP block in the image encoder of the CLIP.