Research map / Image generation / Autoregressive and latent generation
Forensic-Friendly Image Manipulation via Controllable Latent Diffusion
Hanyu Chen, Haiwei Wu, Jinyu Tian, Jianqing LI, Jiantao Zhou
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 FFIM instead of the DDPM baseline for the in-domain forensic model | DDPM editing · AniCOCO, TruFor | F1: .570 -> .700 | Improved |
| post-hoc ReLoc enhancement of edited images | DDPM editing · AniCOCO, TruFor | F1: .570 -> .557 | Hurt |
| ReLoc across localization forensics | DDPM editing · Pixel-level localization (four datasets) | F1 | Mixed |
| FFIM on image-level detectors | DDPM editing · Image-level detection (CNND, UFD, DRCT) | AUC: +14.8%, +13.5%, +27.3% AUC | Improved |
| ReLoc on image-level detectors | DDPM editing · Image-level detection (CNND, UFD, DRCT) | AUC: +5.9%, +11.8%, -6.2% AUC | Mixed |
| add FFIM to the editing process | DDPM editing · User study (20 participants) | satisfaction score: 4.15 vs 4.17, 4.10, 4.21 | No clear effect |
| add controllable projection to explicit guidance (ablation) | FFIM with TruFor guidance · AniCOCO, TruFor | F1: .619 -> .700 | Improved |
| add explicit guidance on top of controllable projection (ablation) | FFIM · AniCOCO, TruFor | F1: .570 -> .700 | Improved |
| use l-infinity norm projection instead of l2 (ablation) | FFIM · AniCOCO, TruFor | F1: .700 -> .900 | Mixed |
| use cosine-similarity projection instead of l2 (ablation) | FFIM · AniCOCO, TruFor | F1: .700 -> .599 | Hurt |
| use l1 instead of l2 norm projection (ablation) | FFIM · AniCOCO, TruFor | F1: .700 -> .700 | No clear effect |
| apply FFIM with DPM denoising (ablation) | DPM · AniCOCO, TruFor | F1: .568 -> .692 | Improved |
Closest papers by meaning
- ForensicHub: A Unified Benchmark & Codebase for All-Domain Fake Image Detection and Localization (NeurIPS 2025 · similarity 0.87)
- Training-Free Image Manipulation Localization Using Diffusion Models (AAAI 2025 · similarity 0.87)
- FakeShield: Explainable Image Forgery Detection and Localization via Multi-modal Large Language Models (ICLR 2025 · similarity 0.87)
- DiffusionGuard: A Robust Defense Against Malicious Diffusion-based Image Editing (ICLR 2025 · similarity 0.87)
- Less Supervision, Better Generalization: Weakly Supervised Fake Region Localization in Diffusion-Edited Images (NeurIPS 2026 · similarity 0.87)
- Learning Discriminative Noise Guidance for Image Forgery Detection and Localization (AAAI 2024 · 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.
