GC-ConsFlow: Leveraging Optical Flow Residuals and Global Context for Robust Deepfake Detection

TL;DR

GC-ConsFlow tackles Deepfake detection by unifying spatial and temporal cues through a dual-stream architecture. The Global Context-Aware Frame Flow (GCAF) with the GGCA module enhances frame-level spatial representations, while the Flow-Gradient Temporal Consistency (FGTC) stream uses optical-flow residuals and gradient features to robustly model temporal inconsistencies amidst natural facial motion. Ablation results show complementarity between streams and the value of reconstructed-frame residuals and HOG features, with GC-ConsFlow achieving state-of-the-art results on FF++ HQ and solid generalization to Celeb-DF under compression. Overall, the method offers a robust, generalizable approach to Deepfake detection with practical potential for deployment in varied video conditions.

Abstract

The rapid development of Deepfake technology has enabled the generation of highly realistic manipulated videos, posing severe social and ethical challenges. Existing Deepfake detection methods primarily focused on either spatial or temporal inconsistencies, often neglecting the interplay between the two or suffering from interference caused by natural facial motions. To address these challenges, we propose the global context consistency flow (GC-ConsFlow), a novel dual-stream framework that effectively integrates spatial and temporal features for robust Deepfake detection. The global grouped context aggregation module (GGCA), integrated into the global context-aware frame flow stream (GCAF), enhances spatial feature extraction by aggregating grouped global context information, enabling the detection of subtle, spatial artifacts within frames. The flow-gradient temporal consistency stream (FGTC), rather than directly modeling the residuals, it is used to improve the robustness of temporal feature extraction against the inconsistency introduced by unnatural facial motion using optical flow residuals and gradient-based features. By combining these two streams, GC-ConsFlow demonstrates the effectiveness and robustness in capturing complementary spatiotemporal forgery traces. Extensive experiments show that GC-ConsFlow outperforms existing state-of-the-art methods in detecting Deepfake videos under various compression scenarios.

Figures (3)

• Figure 1: Overview of the proposed GC-ConsFlow framework for deepfake detection.
• Figure 2: The Global Grouped Context Aggregation (GGCA) is designed to enhance spatial feature extraction by aggregating grouped global context information.
• Figure 3: Comparison of heatmaps generated with and without reconstructed frames on real and forged video frames.