Research map / Image generation / Image restoration
Tackling Structural Hallucination in Image Translation with Local Diffusion
Seunghoi Kim, Chen Jin, Tom Diethe, Matteo Figini, Henry FJ Tregidgo, Asher Mullokandov, Philip A Teare, Daniel 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 |
|---|---|---|---|
| add Local Diffusion (branching and fusion) to DDPM | DDPM · BraTS tumor segmentation | Dice coefficient: 0.194 -> 0.590 | Improved |
| add Local Diffusion to DDPM | DDPM · MVTec AD anomaly detection | accuracy (%): 58.4 -> 85.0 | Improved |
| add Local Diffusion to DDPM | DDPM · MNIST digit classification | accuracy (%): 95.7 -> 98.0 | Improved |
| add Local Diffusion to DDPM | DDPM · BraTS image quality | PSNR: 19.5 -> 21.2 | Improved |
| add DSI (distribution shift inversion) to DDPM | DDPM · BraTS tumor segmentation | Dice coefficient: 0.194 -> 0.100 | Hurt |
| apply Local Diffusion on top of DDIM with fewer steps (0.1T vs 0.5T) (ablation) | DDIM + Ours · MVTec AD anomaly detection | accuracy (%): 66.8 (0.1T) vs 83.2 (0.5T) | Mixed |
| use manual or U-Net OOD segmentation instead of PatchCore (ablation) | Local Diffusion · BraTS | Dice coefficient: 0.590 -> 0.670 (manual), 0.671 (U-Net) | Improved |
| start fusion stage earlier (ablation) | Local Diffusion · MVTec AD | accuracy vs FID | Mixed |
| apply Local Diffusion to images with multiple OOD regions (ablation) | DDPM · BraTS multiple OOD | Dice coefficient: 0.055 -> 0.540 | Improved |
| apply Local Diffusion to MNIST image quality | DDPM · MNIST | PSNR: 20.8 -> 20.9 | No clear effect |
Closest papers by meaning
- Understanding Hallucinations in Diffusion Models through Mode Interpolation (NeurIPS 2024 · similarity 0.91)
- RODS: Robust Optimization Inspired Diffusion Sampling for Detecting and Reducing Hallucination in Generative Models (NeurIPS 2025 · similarity 0.89)
- GHOST: Hallucination-Inducing Image Generation for Multimodal LLMs (ICLR 2026 · similarity 0.88)
- Hallucination-aware Intermediate Representation Editing in Large Vision-Lanugage Models (ICLR 2026 · similarity 0.87)
- Towards Understanding Text Hallucination of Diffusion Models via Local Generation Bias (ICLR 2025 · similarity 0.87)
- ConVis: Contrastive Decoding with Hallucination Visualization for Mitigating Hallucinations in Multimodal Large Language Models (AAAI 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.
