Research map / Detection and segmentation / Open-set and few-shot detection
Detecting and Recovering Sequential DeepFake Manipulation
Rui Shao, Tianxing Wu, Ziwei Liu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| add auto-regressive decoding (ablation) | SeqFakeFormer (ResNet-34) · Seq-DeepFake (facial components) | Adaptive-Acc: 51.71 -> 54.80 | Improved |
| add SECA with auto-regressive decoding (ablation) | SeqFakeFormer (ResNet-34) · Seq-DeepFake (facial components) | Adaptive-Acc: 53.84 -> 54.80 | Improved |
| add SECA without auto-regressive decoding (ablation) | SeqFakeFormer (ResNet-34) · Seq-DeepFake (facial components) | Adaptive-Acc: 52.19 -> 51.71 | Hurt |
| add SECA without auto-regressive decoding (ablation) | SeqFakeFormer (ResNet-50) · Seq-DeepFake (facial components) | Fixed-Acc: 71.22 -> 70.99 | Hurt |
| add SECA (ablation) | SeqFakeFormer (ResNet-34) · Seq-DeepFake (facial attributes) | Adaptive-Acc: 46.56 -> 48.32 | Improved |
| add auto-regressive decoding (ablation) | SeqFakeFormer (ResNet-34) · Seq-DeepFake (facial attributes) | Adaptive-Acc: 47.22 -> 48.32 | Improved |
| add auto-regressive decoding alone (no SECA) (ablation) | SeqFakeFormer (ResNet-34) · Seq-DeepFake (facial attributes) | Fixed-Acc: 66.98 -> 66.70 | Hurt |
| replace Multi-Cls with SeqFakeFormer | Multi-Cls (ResNet-50) · Seq-DeepFake (facial components) | Adaptive-Acc: 50.57 -> 55.30 | Improved |
| replace Multi-Cls with SeqFakeFormer | Multi-Cls (ResNet-50) · Seq-DeepFake (facial attributes) | Adaptive-Acc: 46.00 -> 49.63 | Improved |
| replace DETR with SeqFakeFormer | DETR (ResNet-34) · Seq-DeepFake (facial attributes) | Fixed-Acc: 67.93 -> 67.99 | No clear effect |
| replace Two-Stream deepfake detector with SeqFakeFormer | Two-Stream · Seq-DeepFake (facial components) | Adaptive-Acc: 53.89 -> 55.30 | Improved |
| evaluate with adaptive length instead of fixed length | SeqFakeFormer (ResNet-50) · Seq-DeepFake (facial components) | accuracy: 72.65 -> 55.30 | Hurt |
Closest papers by meaning
- DF40: Toward Next-Generation Deepfake Detection (NeurIPS 2024 · similarity 0.88)
- DiffusionFake: Enhancing Generalization in Deepfake Detection via Guided Stable Diffusion (NeurIPS 2024 · similarity 0.88)
- Rethinking Vision-Language Model in Face Forensics: Multi-Modal Interpretable Forged Face Detector (CVPR 2025 · similarity 0.87)
- From Specificity to Generality: Revisiting Generalizable Artifacts in Detecting Face Deepfakes (NeurIPS 2025 · similarity 0.87)
- Standing on the Shoulders of Giants: Reprogramming Visual-Language Model for General Deepfake Detection (AAAI 2025 · similarity 0.87)
- Towards More General Video-based Deepfake Detection through Facial Component Guided Adaptation for Foundation Model (CVPR 2025 · similarity 0.87)
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.
