Research map / Image generation / Text-to-image diffusion
Detect Any AI-Counterfeited Text Image
Chenfan Qu, Yiwu Zhong, Xuekang Zhu, Junchi Li, Changjiang Jiang, Jian liu, Lianwen Jin
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 internal patch shuffling on non-text images (ablation) | DS-Net · DanceText | instance-level F1: 51.6 -> 53.9 | Improved |
| add external patch shuffling (ablation) | DS-Net · DanceText | instance-level F1: 50.5 -> 53.9 | Improved |
| add non-text decoupling loss Ldnt (ablation) | DS-Net · DanceText | instance-level F1: 47.9 -> 53.9 | Improved |
| add text decoupling loss Ldt (ablation) | DS-Net · DanceText | instance-level F1: 51.1 -> 53.9 | Improved |
| add domain alignment loss Lda (ablation) | DS-Net · DanceText | instance-level F1: 49.0 -> 53.9 | Improved |
| add Artifact Decouple Network with its three losses (ablation) | DS-Net · DanceText | instance-level F1: 47.3 -> 53.9 | Improved |
| add Global Forensic Query (synergy decoder) (ablation) | DS-Net · DanceText | instance-level F1: 51.8 -> 53.9 | Improved |
| add Global Forensic Query (synergy decoder) (ablation) | DS-Net · T-IC13 | F1: 95.3 -> 95.5 | Improved |
| add all proposed modules to DINO baseline (ablation) | DINO · DanceText | balanced accuracy: 74.2 -> 77.4 | Improved |
| change generator (cross-generator test) for all detectors | existing detectors · DanceText-CG vs Test | balanced accuracy: 92.6 -> 75.6 (CRCNN-DAF) | Hurt |
| change language of test text | existing detectors · DanceText-CL vs Test | instance F1: 78.5 -> 64.0 (CRCNN-DAF) | Hurt |
Closest papers by meaning
- Revisiting Tampered Scene Text Detection in the Era of Generative AI (AAAI 2025 · similarity 0.88)
- Spot the Fake: Large Multimodal Model-Based Synthetic Image Detection with Artifact Explanation (NeurIPS 2025 · similarity 0.87)
- FakeShield: Explainable Image Forgery Detection and Localization via Multi-modal Large Language Models (ICLR 2025 · similarity 0.86)
- FakeInversion: Learning to Detect Images from Unseen Text-to-Image Models by Inverting Stable Diffusion (CVPR 2024 · similarity 0.85)
- TextShield-R1: Reinforced Reasoning for Tampered Text Detection (AAAI 2026 · similarity 0.85)
- Community Forensics: Using Thousands of Generators to Train Fake Image Detectors (CVPR 2025 · similarity 0.85)
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.
