R-FLoRA: Residual-Statistic-Gated Low-Rank Adaptation for Single-Image Face Morphing Attack Detection

TL;DR

This paper introduces R-FLoRA, a novel single-image face morphing attack detection framework that combines residual statistics with vision transformers, achieving high accuracy, generalisation, and real-time efficiency.

Contribution

The paper proposes R-FLoRA, a new method integrating residual-statistic gating and low-rank adapters with vision transformers for improved face morphing attack detection.

Findings

Outperforms nine recent S-MAD algorithms in accuracy

Demonstrates strong cross-domain generalisation across four datasets

Achieves real-time detection with minimal trainable parameters

Abstract

Face morphing attacks pose a substantial risk to the reliability of face recognition systems used in passport issuance, border control, and digital identity verification. Detecting morphing attacks from a single facial image remains challenging owing to the lack of a trusted reference and the diversity of attack generation methods. This paper presents a new Single-Image Face Morphing Attack Detection (S-MAD) framework that integrates high-frequency Laplacian residual statistics with representations from a frozen, foundation-scale vision transformer. The approach employs residual-statistic-gated low-rank adapters (R-FLoRA) and feature-wise residual fusion (Res-FiLM) to enhance sensitivity to local morphing artefacts while preserving the semantic context of the backbone. A novel residual-contrastive alignment loss further regularises the fused token space, improving discrimination under unseen morphing conditions. Comprehensive experiments on four ICAO-compliant datasets, encompassing seven morph generation techniques, demonstrate that the proposed method consistently surpasses nine recent state-of-the-art S-MAD algorithms in detection accuracy and cross-domain (or dataset) generalisation. With a frozen backbone and minimal trainable parameters, the model achieves real-time efficiency and interpretability, making it suitable for real-life scenarios in biometric verification systems.