🤖 AI Summary
This work addresses the limited robustness of deepfake detection in real-world scenarios, where diverse forgery techniques and image degradations obscure manipulation traces. To this end, the authors propose a local–global decoupled modeling framework: a global branch leverages multi-scale heterogeneous vision foundation models to capture holistic anomalies, while a local branch employs multiple instance learning with top-k aggregation to focus on suspicious regions. Dual-level supervision is introduced to preserve discriminative power, and predictions from both branches are fused in logit space. By combining heterogeneous backbones and multi-granularity supervision, the method effectively decorrelates errors and mitigates the dilution of forgery signals by authentic image regions. The approach achieved second place in the NTIRE 2026 Robust Deepfake Detection Challenge and demonstrates strong generalization and robustness across multiple public benchmarks.
📝 Abstract
Robust deepfake detection in the wild remains challenging due to the ever-growing variety of manipulation techniques and uncontrolled real-world degradations. Forensic cues for deepfake detection reside at two complementary levels: global-level anomalies in semantics and statistics that require holistic image understanding, and local-level forgery traces concentrated in manipulated regions that are easily diluted by global averaging. Since no single backbone or input scale can effectively cover both levels, we propose LOGER, a LOcal--Global Ensemble framework for Robust deepfake detection. The global branch employs heterogeneous vision foundation model backbones at multiple resolutions to capture holistic anomalies with diverse visual priors. The local branch performs patch-level modeling with a Multiple Instance Learning top-$k$ aggregation strategy that selectively pools only the most suspicious regions, mitigating evidence dilution caused by the dominance of normal patches; dual-level supervision at both the aggregated image level and individual patch level keeps local responses discriminative. Because the two branches differ in both granularity and backbone, their errors are largely decorrelated, a property that logit-space fusion exploits for more robust prediction. LOGER achieves 2nd place in the NTIRE 2026 Robust Deepfake Detection Challenge, and further evaluation on multiple public benchmarks confirms its strong robustness and generalization across diverse manipulation methods and real-world degradation conditions.