Towards Zero-Shot Domain Generalization for ID Cards Presentation Attack Detection
2026-08-17 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors address the problem of detecting fake national ID cards when there are very few real card samples available, especially from different countries. They use a special neural network setup that can learn from just four real cards per type, and train it to recognize common signs of fakes across various card domains. Tested on a large, diverse dataset, their method works better than traditional approaches, even with limited data. This helps make ID verification more accurate and easier to scale without needing lots of personal data.
Presentation-Attack Detection (PAD)Prototypical NetworkEfficientNet-V2Episodic TrainingEqual Error Rate (EER)National ID CardsCross-domain GeneralizationBiometricsRemote OnboardingZero-shot Learning
Authors
Mario Nieto-Hidalgo, Juan M. Espin, Juan E. Tapia
Abstract
Presentation-Attack Detection (PAD) for national ID cards is limited by the lack of publicly available genuine samples, making it difficult for systems to generalize across countries. This paper introduces two main innovations: (1) a Prototypical Network head using an EfficientNet-V2-b0 backbone that requires only four genuine samples per class to create reliable prototypes; and (2) an episodic training regime that keeps PAD classes fixed while varying the card domain, allowing the network to learn universal attack cues. Evaluated on a large multi-country dataset and the public DLC-2021 benchmark, this method achieves an average Equal Error Rate of around 9\%, outperforming conventional softmax and CLIP zero-shot baselines even with data from a single source country. This approach provides accurate, privacy-preserving PAD while minimizing data collection, facilitating scalable cross-jurisdictional remote onboarding.