Research map / Vision-language and multimodal / Multimodal LLMs
UC-VLM: Consistency-Driven Learning for AI-Generated Image Detection with Vision-Language Large Models
Lei Tan, Shuwei Li, Mohan Kankanhalli, Robby T. Tan
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 instruction optimization to frozen backbone (ablation) | Qwen2.5-VL-7B · Chameleon | accuracy: 55.5 -> 68.3 | Improved |
| Add instruction optimization to frozen backbone (ablation) | Qwen2.5-VL-7B · GenImage | accuracy: 51.9 -> 74.2 | Improved |
| Add visual adaptation (local crop + patch shuffle) to visual fine-tuning (ablation) | Qwen2.5-VL-7B · GenImage | accuracy: 76.8 -> 91.7 | Improved |
| Add visual adaptation to visual fine-tuning (ablation) | Qwen2.5-VL-7B · Chameleon | accuracy: 50.9 -> 55.4 | Improved |
| Add label-conditioned regeneration to LLM fine-tuning (ablation) | Qwen2.5-VL-7B · GenImage | accuracy: 80.3 -> 82.4 | Improved |
| Add label-conditioned regeneration to LLM fine-tuning (ablation) | Qwen2.5-VL-7B · Chameleon | accuracy: 71.0 -> 65.1 | Hurt |
| Naively fine-tune both vision and language (ablation) | Qwen2.5-VL-7B · Chameleon | accuracy: 71.0 -> 52.0 | Hurt |
| Combine all three components (full UC-VLM) vs best single component (ablation) | Qwen2.5-VL-7B · GenImage | accuracy: 91.7 -> 96.1 | Improved |
| Use structured prompt with explanation (P-3) vs plain QA (P-1) (ablation) | Qwen2.5-VL-7B · Chameleon | accuracy: 55.5 -> 63.6 | Improved |
| Fine-tune vision encoder only (ablation) | Qwen2.5-VL-7B · Chameleon | accuracy: 55.5 -> 50.9 | Hurt |
Closest papers by meaning
- Roboflow100-VL: A Multi-Domain Object Detection Benchmark for Vision-Language Models (NeurIPS 2025 · similarity 0.88)
- Locate-Then-Examine: Grounded Region Reasoning Improves Detection of AI-Generated Images (CVPR 2026 · similarity 0.88)
- Identity-Aware Vision-Language Model for Explainable Face Forgery Detection (AAAI 2026 · similarity 0.87)
- ROD-MLLM: Towards More Reliable Object Detection in Multimodal Large Language Models (CVPR 2025 · similarity 0.87)
- VadCLIP: Adapting Vision-Language Models for Weakly Supervised Video Anomaly Detection (AAAI 2024 · similarity 0.87)
- VERA: Explainable Video Anomaly Detection via Verbalized Learning of Vision-Language Models (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.
