Research map / Video / Video segmentation and tracking
FInfer: Frame Inference-Based Deepfake Detection for High-Visual-Quality Videos
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| predict future facial representations from current facial representations | FInfer · Celeb-DF | AUC (%): 93.30 | Helped |
| predict future facial representations from current facial representations | FInfer · WildDeepfake | ACC (%): 75.88 | Helped |
| predict future facial representations from current facial representations | FInfer · DFDC-preview | ACC (%): 80.39 | Helped |
| use frame inference rather than direct frame-feature extraction | FInfer · Celeb-DF (trained on FF++) | AUC (%): 70.60; SPSL 76.88 | Mixed |
| add the Gaussian-Laplace pyramid block (ablation) | FInfer · Celeb-DF | ACC (%): 86.68 -> 90.47 | Helped |
| add the Gaussian-Laplace pyramid block (ablation) | FInfer · WildDeepfake | ACC (%): 75.74 -> 75.88 | No clear effect |
| add the Gaussian-Laplace pyramid block (ablation) | FInfer · DFDC-preview | ACC (%): 77.17 -> 80.39 | Helped |
| use 30 source frames and 20 target frames (ablation) | FInfer · Celeb-DF | ACC (%): 90.47 | Helped |
| increase target length to 30 frames (s=30) (ablation) | FInfer · Celeb-DF | ACC (%): 90.47 -> 86.72 | Hurt |
| increase source length from 30 to 40 frames (t=20) (ablation) | FInfer · Celeb-DF | ACC (%): 90.47 -> 88.48 | Hurt |
| train with longer target sequences (ablation) | FInfer · Celeb-DF | ACC (%): 89.26; 90.47; 86.72; 88.87 | Mixed |
Closest papers by meaning
- Towards More General Video-based Deepfake Detection through Facial Component Guided Adaptation for Foundation Model (CVPR 2025 · similarity 0.87)
- Towards a Universal Synthetic Video Detector: From Face or Background Manipulations to Fully AI-Generated Content (CVPR 2025 · similarity 0.87)
- DF40: Toward Next-Generation Deepfake Detection (NeurIPS 2024 · similarity 0.86)
- Generalizing Deepfake Video Detection with Plug-and-Play: Video-Level Blending and Spatiotemporal Adapter Tuning (CVPR 2025 · similarity 0.86)
- Exploiting Style Latent Flows for Generalizing Deepfake Video Detection (CVPR 2024 · similarity 0.86)
- AVFF: Audio-Visual Feature Fusion for Video Deepfake Detection (CVPR 2024 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.