Research map / Vision-language and multimodal / Multimodal LLMs
Semantic Discrepancy-aware Detector for Image Forgery Identification
Ziye Wang, Minghang Yu, Chunyan Xu, Zhen Cui
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 Feature Enhancer (ablation) | CFDL only (no STS) · UnivFD dataset | mean accuracy: 81.64 -> 90.17 | Improved |
| add Feature Enhancer (ablation) | CFDL only (no STS) · UnivFD dataset | mean average precision: 97.37 -> 97.41 | No clear effect |
| add semantic token sampling (STS) (ablation) | CFDL · UnivFD dataset | mean accuracy: 81.64 -> 89.98 | Improved |
| add STS on top of CFDL + Feature Enhancer (ablation) | SDD · UnivFD dataset | mean accuracy: 90.17 -> 93.61 | Improved |
| add STS on top of CFDL + Feature Enhancer (ablation) | SDD · UnivFD dataset | mean average precision: 97.41 -> 98.52 | Improved |
| add Feature Enhancer on top of STS + CFDL (ablation) | SDD · UnivFD dataset | mean accuracy: 89.98 -> 93.61 | Improved |
| add Feature Enhancer on top of STS + CFDL (ablation) | SDD · UnivFD dataset | mean average precision: 97.39 -> 98.52 | Improved |
| use CLIP ViT-L/14 backbone instead of BLIP ViT-L/16 (ablation) | SDD · UnivFD dataset | mean average precision: 95.61 -> 98.52 | Improved |
| use exponential inverse adaptive weight instead of |x| or x^2 (ablation) | SDD · UnivFD dataset | mean AP / accuracy | Improved |
| raise token sampling rate from 1/500 to 1/1000 (ablation) | SDD · UnivFD dataset | mean accuracy | Improved |
| SDD versus FatFormer (best baseline) | SDD · UnivFD dataset | mean AP / accuracy: +0.34% AP, +2.75% accuracy | Improved |
| SDD versus NPR (best baseline) | SDD · SynRIS | mean AUROC: 90.3 -> 95.1 | Improved |
Closest papers by meaning
- Identity-Aware Vision-Language Model for Explainable Face Forgery Detection (AAAI 2026 · similarity 0.90)
- Reinforced Multi-teacher Knowledge Distillation for Efficient General Image Forgery Detection and Localization (AAAI 2025 · similarity 0.88)
- Learning Discriminative Noise Guidance for Image Forgery Detection and Localization (AAAI 2024 · similarity 0.88)
- Forensics-Bench: A Comprehensive Forgery Detection Benchmark Suite for Large Vision Language Models (CVPR 2025 · similarity 0.87)
- Unleashing the Potential of Consistency Learning for Detecting and Grounding Multi-Modal Media Manipulation (CVPR 2025 · similarity 0.86)
- FakeShield: Explainable Image Forgery Detection and Localization via Multi-modal Large Language Models (ICLR 2025 · 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.
