Research map / Label-efficient and robust learning / Self-supervised and contrastive
Fighting Fake News: Image Splice Detection via Learned Self-Consistency
Minyoung Huh, Andrew Liu, Andrew Owens, Alexei A. Efros
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 |
|---|---|---|---|
| use EXIF metadata consistency instead of directly predicting same-image consistency (ablation) | patch consistency model · Splice detection, Carvalho | mAP: 0.75 -> 0.87 | Improved |
| use EXIF metadata consistency instead of same-image consistency (ablation) | patch consistency model · Splice localization, Hays | mAP: 0.21 -> 0.48 | Improved |
| use EXIF metadata consistency instead of same-image consistency (ablation) | patch consistency model · Splice localization, Carvalho | mAP: 0.36 -> 0.51 | Improved |
| use EXIF metadata consistency instead of same-image consistency (ablation) | patch consistency model · Splice localization, Realistic Tampering | mAP: 0.21 -> 0.20 | No clear effect |
| train longer on EXIF vs Image consistency (ablation) | Image-Consistency · splice localization | gap between models | Mixed |
| use self-supervised EXIF consistency instead of a supervised FCN trained on spliced images | patch consistency model · Splice localization, Columbia | mAP: 0.60 -> 0.91 | Improved |
| use EXIF consistency instead of camera classification (ablation) | patch consistency model · Splice detection, Realistic Tampering | mAP: 0.15 -> 0.55 | Improved |
| use EXIF consistency instead of E-MFCN | patch consistency model · Splice localization vs Salloum et al., Columbia (full) | F1: 0.61 -> 0.88 | Improved |
| use EXIF consistency instead of E-MFCN | patch consistency model · Splice localization vs Salloum et al., Carvalho (full) | F1: 0.48 -> 0.52 | Improved |
| use EXIF consistency instead of CFA | patch consistency model · Splice localization, Realistic Tampering | mAP: 0.40 -> 0.20 | Hurt |
| add post-processing consistency (re-JPEG, blur, resize) tasks (ablation) | EXIF-Consistency · Realistic Tampering / copy-move | detection of post-processing cues | Improved |
Closest papers by meaning
- ForensicHub: A Unified Benchmark & Codebase for All-Domain Fake Image Detection and Localization (NeurIPS 2025 · similarity 0.86)
- Forensic Self-Descriptions Are All You Need for Zero-Shot Detection, Open-Set Source Attribution, and Clustering of AI-generated Images (CVPR 2025 · similarity 0.85)
- M²RL-Net: Multi-View and Multi-Level Relation Learning Network for Weakly-Supervised Image Forgery Detection (AAAI 2025 · similarity 0.85)
- Learning Discriminative Noise Guidance for Image Forgery Detection and Localization (AAAI 2024 · similarity 0.85)
- A Bias-Free Training Paradigm for More General AI-generated Image Detection (CVPR 2025 · similarity 0.85)
- FakeShield: Explainable Image Forgery Detection and Localization via Multi-modal Large Language Models (ICLR 2025 · similarity 0.84)
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.
