Towards Robust DeepFake Detection under Unstable Face Sequences: Adaptive Sparse Graph Embedding with Order-Free Representation and Explicit Laplacian Spectral Prior

2025-12-08

AI Summary

Existing DeepFake detectors suffer severe performance degradation when applied to real-world videos corrupted by compression artifacts, occlusions, or adversarial attacks—conditions that cause missing, misordered, or heavily degraded face sequences. To address this, we propose a robust detection framework centered on an order-agnostic temporal graph embedding mechanism: it constructs adaptive sparse graphs based on semantic similarity and employs a two-level sparsification strategy; further, it explicitly incorporates a graph Laplacian spectral prior in the frequency domain to enhance high-pass responses to tampering traces. Crucially, the method requires only clean face training data and reliably models cross-frame semantic consistency even under severe disorder and noise. Evaluated on FF++, Celeb-DFv2, and DFDC, it achieves state-of-the-art performance—particularly under extreme corruptions—demonstrating superior robustness and generalization.

Technology Category

• Computer Vision: Adversarial Attacks & Robustness
• Machine Learning: Adversarial Learning & Robustness
• Natural Language Processing: Safety and Robustness

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Ensuring the authenticity of video content remains challenging as DeepFake generation becomes increasingly realistic and robust against detection. Most existing detectors implicitly assume temporally consistent and clean facial sequences, an assumption that rarely holds in real-world scenarios where compression artifacts, occlusions, and adversarial attacks destabilize face detection and often lead to invalid or misdetected faces. To address these challenges, we propose a Laplacian-Regularized Graph Convolutional Network (LR-GCN) that robustly detects DeepFakes from noisy or unordered face sequences, while being trained only on clean facial data. Our method constructs an Order-Free Temporal Graph Embedding (OF-TGE) that organizes frame-wise CNN features into an adaptive sparse graph based on semantic affinities. Unlike traditional methods constrained by strict temporal continuity, OF-TGE captures intrinsic feature consistency across frames, making it resilient to shuffled, missing, or heavily corrupted inputs. We further impose a dual-level sparsity mechanism on both graph structure and node features to suppress the influence of invalid faces. Crucially, we introduce an explicit Graph Laplacian Spectral Prior that acts as a high-pass operator in the graph spectral domain, highlighting structural anomalies and forgery artifacts, which are then consolidated by a low-pass GCN aggregation. This sequential design effectively realizes a task-driven spectral band-pass mechanism that suppresses background information and random noise while preserving manipulation cues. Extensive experiments on FF++, Celeb-DFv2, and DFDC demonstrate that LR-GCN achieves state-of-the-art performance and significantly improved robustness under severe global and local disruptions, including missing faces, occlusions, and adversarially perturbed face detections.

Problem

• Detects DeepFakes in noisy or unordered face sequences
• Handles invalid faces from compression, occlusions, and attacks
• Uses graph spectral prior to highlight forgery artifacts

Innovation

• Adaptive sparse graph embedding with order-free representation
• Dual-level sparsity mechanism for graph structure and features
• Explicit Graph Laplacian Spectral Prior as high-pass operator