Micro-Expression-Aware Avatar Fingerprinting via Inter-Frame Feature Differencing

TL;DR

This paper introduces a novel avatar fingerprinting method that leverages inter-frame feature differencing on raw video frames with a micro-expression-aware backbone, enabling end-to-end optimization and improved identity verification.

Contribution

The proposed system eliminates external preprocessing, uses a deep feature differencing approach with a specialized backbone, and demonstrates superior performance in avatar driver identification.

Findings

Achieves an AUC of 0.877 on NVFAIR dataset.

Outperforms landmark-based baseline on most cross-generator pairs.

Temporal motion features are the primary discriminative factor.

Abstract

Avatar fingerprinting, i.e., verifying who drives a synthetic talking-head video rather than whether it is real, is a critical safeguard for authorized use of face-reenactment technology. Existing methods rely on a fixed, non-differentiable landmark extraction stage that prevents the fingerprinting model from being optimized end-to-end from raw pixels. We propose a preprocessing-free system built on a micro-expression-aware backbone operating on raw video frames, with inter-frame feature differencing as the core design principle: consecutive feature maps are subtracted in the learned deep feature space, so that temporally stable appearance dimensions contribute zero to the output while driver-specific motion dynamics are preserved. A controlled ablation on NVFAIR confirms that temporal motion accounts for the large majority of discriminative performance, and that raw appearance features actively degrade identity separation. Both the choice of backbone and the differencing principle are essential: differencing alone is insufficient when applied to a generic encoder, as appearance-dominated features collapse to near-identical representations across adjacent frames, while the micro-expression-aware F5C backbone retains measurable motion variation that the differencing operation can exploit. Without any external preprocessing, our model achieves an overall AUC of 0.877 on NVFAIR and matches or exceeds the landmark-based baseline on the majority of cross-generator pairs.