Guided and Fused: Efficient Frozen CLIP-ViT with Feature Guidance and Multi-Stage Feature Fusion for Generalizable Deepfake Detection

Re-assign community