Research map / Image generation / Image restoration
EndoIR: Degradation-Agnostic All-in-One Endoscopic Image Restoration via Noise-Aware Routing Diffusion
Tong Chen, Xinyu Ma, Long Bai, Wenyang Wang, Yue Sun, Luping Zhou
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 Dual-Stream Encoder (ablation) | EndoIR baseline · SegSTRONG-C average | PSNR: 30.39 -> 31.96 | Improved |
| add Task Adaptive Embedding (ablation) | EndoIR baseline · SegSTRONG-C average | PSNR: 30.39 -> 31.40 | Improved |
| add Rectified Fusion Block (ablation) | EndoIR baseline · SegSTRONG-C average | PSNR: 30.39 -> 31.56 | Improved |
| add RFB on top of TAE+DSE (ablation) | EndoIR · SegSTRONG-C average | PSNR: 32.20 -> 32.36 | Improved |
| add RFB on top of TAE only (ablation) | TAE · SegSTRONG-C average | PSNR: 31.40 -> 31.51 | Improved |
| use both spatial and frequency embeddings in Dual-Domain Prompter (ablation) | Dual-Domain Prompter · SegSTRONG-C average | PSNR: 31.19 -> 32.36 | Improved |
| use FFT instead of wavelet transform in prompter (ablation) | Dual-Domain Prompter · SegSTRONG-C average | PSNR: 31.96 -> 32.36 | Improved |
| reduce NARB channel selection ratio from 1 to 0.5 (ablation) | Noise-Aware Routing Block · SegSTRONG-C average | SSIM: 87.05 -> 87.64 SSIM, 32.36 -> 32.23 PSNR | Mixed |
| select 0.75 of channels instead of 0.5 (ablation) | Noise-Aware Routing Block · SegSTRONG-C average | PSNR: 32.23 -> 32.02 | Hurt |
| NARB with gamma 0.5 for efficiency (ablation) | EndoIR · SegSTRONG-C | Params (M) and FPS: 28.2 -> 21.3 M params, 8.14 -> 11.18 FPS | Improved |
| EndoIR vs best baseline on CEC | EndoIR · CEC illumination correction | PSNR: 32.27 -> 32.65 | Improved |
| restore images with EndoIR before segmentation | DeepLabV3+ segmentation · SegSTRONG-C downstream | Dice: 95.47 -> 96.22 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- PromptIR — PromptIR: Prompting for All-in-One Image Restoration (NeurIPS 2023) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Benchmarking Endoscopic Surgical Image Restoration and Beyond (CVPR 2026 · similarity 0.89)
- EndoBench: A Comprehensive Evaluation of Multi-Modal Large Language Models for Endoscopy Analysis (NeurIPS 2025 · similarity 0.85)
- Multi-view Masked Contrastive Representation Learning for Endoscopic Video Analysis (NeurIPS 2024 · similarity 0.83)
- Degradation-Aware Feature Perturbation for All-in-One Image Restoration (CVPR 2025 · similarity 0.83)
- A Unified Degradation-Robust Approach to SSL and UDA for 3D Medical Images (AAAI 2025 · similarity 0.82)
- Resfusion: Denoising Diffusion Probabilistic Models for Image Restoration Based on Prior Residual Noise (NeurIPS 2024 · similarity 0.82)
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.
