Video Transformer for Remote Identity Document Hologram Detection

Authors: Joris Voerman, Nicolas Sidere, Jean-Christophe Burie

Published: 2026-07-13 11:24:54+00:00

Comment: Accepted to SMC2026 conference, 6 pages, 5 figures, 5 tables

AI Summary

This paper proposes a Remote Identity Document Verification System (RIDVS) featuring a video transformer-based approach (ViTransHolo) for detecting holograms in ID documents captured by smartphones. The system focuses on underutilized hologram security features which are difficult to counterfeit and detectable through smartphone video. The proposed method significantly outperforms existing state-of-the-art techniques, achieving improved accuracy and recall on both public and private datasets.

Abstract

Remote identity authentification using Identification Documents has been a major challenge for several years. DeepFakes advent and the development of AI-guided tools helps fraudsters creating counterfeit ID Documents. Ensuring the authenticity of ID Documents has become a real clue in the seurization of remote authentification. This need is all the more pressing given the increasing digitization of administrative and transactional processes. To ensure widespread accessibility, the system should rely solely on video captured via mobile devices. In this specific context, confirming the authenticity of ID is a real challenge as many security features needs specific device like infrared sensor for instance. Among underutilized but promising security features, holographic printings hold a special place. Difficult to counterfeit, they produce distinctive visual effects according enlightment, making them both detectable in a video captured by a smartphone camera and difficult to imitate. In this paper, we propose a Remote Identity Document Verification System (RIDVS) and an approach based on a video transformer for detecting holograms in simple videos captured by smartphones. Our system is designed for a smartphone-based capture process, followed by a server-side verification. The hologram detection method builds on a robust model previously validated in a related research domain. We demonstrate that it outperforms existing SotA methods, achieving near-perfect accuracy even when trained on medium- to small-sized datasets. In particular, we report improvements of +26.86\\% in Recall and +17.93\\% in accuracy over the best MIDV-Holo baseline. This study includes several experiments that evaluate the model adaptation to frugality, both for training samples and computational resources.

Key findings

ViTransHolo achieved significant improvements over the best MIDV-Holo baseline, with +26.86% in Recall and +17.93% in accuracy. The model demonstrated strong generalization to real-world data (IML3 dataset) and unseen fraud types, maintaining high performance even with limited training data and reduced model depth. Frugality experiments showed that high performance is maintained with limited training data and reduced model depth, at only a marginal accuracy cost.

Approach

The proposed ViTransHolo model is based on a video transformer architecture, an adaptation of the ViTransPAD model. It processes input videos by selecting 50 equally spaced frames, crops the ID from the background, and resizes it. Features are extracted using an EfficientNet convolutional encoder, followed by six transformer layers with Multi-Scale Multi-Head Self-Attention. A Multi-Layer Perceptron performs frame-level classification, with video-level predictions derived by averaging frame-level outputs.

Datasets

MIDV-Holo Dataset, IML3 Dataset

Model(s)

ViTransHolo (adapted from ViTransPAD), EfficientNet (as convolutional encoder)

Author countries

France

← Previous