通过三路融合提升深度伪造检测鲁棒性，精准捕捉细微伪造痕迹。
ForensicFlow: A Tri-Modal Adaptive Network for Robust Deepfake Detection
- 融合全局视觉、纹理异常和频谱噪声三类证据，动态加权决策。
- 在CelebDF(v2)上达AUC 0.9752，F1 0.9408，优于单一模型。
- 适合应对复杂伪造场景，可解释性强，适配安全与内容审核需求。
现代深度伪造通过留下细微的领域特异性伪影逃避检测，单分支网络难以捕捉。ForensicFlow通过融合三个取证维度解决此问题：全局视觉不一致（使用ConvNeXt-tiny）、细粒度纹理异常（使用Swin Transformer-tiny）以及频谱噪声模式（使用带通道注意力的CNN）。基于注意力的时序池化动态优先处理高证据帧，自适应融合机制根据伪造类型调整各分支权重。在CelebDF(v2)上用焦点损失训练，模型取得AUC 0.9752、F1 0.9408、准确率0.9208，显著优于单流检测器。消融实验验证了分支协同效应，Grad-CAM可视化证实模型聚焦于真实篡改区域（如面部边界）。该多域融合策略有效提升了对日益复杂的伪造手段的鲁棒性。
原文摘要 · Abstract (English)
Modern deepfakes evade detection by leaving subtle, domain-speci c artifacts that single branch networks miss. ForensicFlow addresses this by fusing evidence across three forensic dimensions: global visual inconsistencies (via ConvNeXt-tiny), ne-grained texture anomalies (via Swin Transformer-tiny), and spectral noise patterns (via CNN with channel attention). Our attention-based temporal pooling dynamically prioritizes high-evidence frames, while adaptive fusion weights each branch according to forgery type. Trained on CelebDF(v2) with Focal Loss, the model achieves AUC 0.9752, F1 0.9408, and accuracy 0.9208 out performing single-stream detectors. Ablation studies con rm branch synergy, and Grad-CAM visualizations validate focus on genuine manipulation regions (e.g., facial boundaries). This multi-domain fusion strategy establishes robustness against increasingly sophisticated forgeries.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。