Research map / Image generation / Image restoration
HalluGen: Synthesizing Realistic and Controllable Hallucinations for Evaluating Image Restoration
Seunghoi Kim, Henry F. J. Tregidgo, Chen Jin, Matteo Figini, Daniel C. Alexander
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 HalluGen instead of vanilla DPS to generate outputs | diffusion restoration · HCP hallucinated regions | segmentation IoU: 0.861 -> 0.363 | Improved |
| swap feature extractor (MedSAM, SAM, DINOv3) (ablation) | HalluGen · HCP hallucinated regions | segmentation IoU: 0.363 / 0.367 / 0.362 | No clear effect |
| increase gradient ascent strength gamma (ablation) | HalluGen intrinsic · HCP | squared error in hallucinated regions: 718 -> 767 | Improved |
| increase number of patches from 1 to 5 (ablation) | HalluGen intrinsic · HCP | total squared error: 150 -> 810 | Improved |
| increase patch size 16x16 to 64x64 (ablation) | HalluGen intrinsic · HCP | squared error: 488 -> 9000 | Improved |
| use SHAFE soft-attention feature metric instead of LPIPS | hallucination metric · Real restoration outputs | AUC: 0.42 -> 0.79 (SHAFE-DINOv3) | Improved |
| use multi-scale SSIM instead of SSIM | pixel metric · intrinsic hallucination | AUC-ROC: 0.52 -> 0.79 | Improved |
| HalluGen vs DPS on SR | HalluGen · HCP-SR x6 | IoU: 0.75 -> 0.62 | Improved |
Closest papers by meaning
- Interpreting and Editing Vision-Language Representations to Mitigate Hallucinations (ICLR 2025 · similarity 0.91)
- Generate, but Verify: Reducing Hallucination in Vision-Language Models with Retrospective Resampling (NeurIPS 2025 · similarity 0.91)
- SAVER: Mitigating Hallucinations in Large Vision-Language Models via Style-Aware Visual Early Revision (AAAI 2026 · similarity 0.90)
- Looks Too Good To Be True: An Information-Theoretic Analysis of Hallucinations in Generative Restoration Models (NeurIPS 2024 · similarity 0.90)
- HaloProbe: Bayesian Detection and Mitigation of Object Hallucinations in Vision-Language Models (ICML 2026 · similarity 0.90)
- Prefill-Time Intervention for Mitigating Hallucination in Large Vision-Language Models (CVPR 2026 · similarity 0.90)
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.
