Research map / Image generation / Image restoration
ALLNet: Multi-task Dense Prediction for Degraded Images
Weiran Wang, Jialing Wu, Yaqi Chang, Gang He, Li Xu, Chang Wu, Yunsong Li
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 MaE restoration experts to MTL baseline (ablation) | ALLNet · Degraded NYUD-v2 | Seg. mIoU: 48.69 -> 53.28 | Improved |
| add MaE to MTL baseline (ablation) | ALLNet · Degraded NYUD-v2 | Depth RMSE: 0.5905 -> 0.5567 | Improved |
| add CTI cross-task interaction unit (ablation) | ALLNet (MaE) · Degraded NYUD-v2 | Seg. mIoU: 53.28 -> 54.75 | Improved |
| add AFU adaptive fusion unit (ablation) | ALLNet (MaE+CTI) · Degraded NYUD-v2 | Depth RMSE: 0.5311 -> 0.5141 | Improved |
| add global task tokens θt (ablation) | ALLNet · Degraded NYUD-v2 | Delta_MTL (%): 3.07 -> 3.87 | Improved |
| add global task tokens θt (ablation) | ALLNet · Degraded NYUD-v2 | Edge odsF: 79.00 -> 79.13 | No clear effect |
| MTL baseline versus single-task learning (ablation) | shared encoder with multiple decoders · Degraded NYUD-v2 | Delta_MTL (%): -6.07 | Hurt |
| one-piece unified network instead of two-stage AdaIR restoration then MLoRE | ALLNet · Degraded NYUD-v2 | Seg. mIoU: 52.43 -> 55.41 | Improved |
| one-piece unified network instead of two-stage AdaIR restoration then MLoRE | ALLNet · Degraded PASCAL-Context | Normal mErr: 14.63 -> 13.92 | Improved |
| learned complexity-bias routing instead of manually fixing one expert (ablation) | MaE · Rain restoration | PSNR: 22.34 -> 28.19 | Improved |
| load-balance routing instead of complexity bias (ablation) | MaE · Haze restoration | PSNR: 31.05 -> 31.64 | Hurt |
Closest papers by meaning
- Debiased All-in-one Image Restoration with Task Uncertainty Regularization (AAAI 2025 · similarity 0.87)
- Controlling Vision-Language Models for Multi-Task Image Restoration (ICLR 2024 · similarity 0.86)
- Learning Where and What to Restore for Composite Image Restoration (NeurIPS 2026 · similarity 0.86)
- Complexity Experts are Task-Discriminative Learners for Any Image Restoration (CVPR 2025 · similarity 0.86)
- Continuous Expert Assembly: Instance-Conditioned Low-Rank Residuals for All-in-One Image Restoration (NeurIPS 2026 · similarity 0.85)
- Vision-Language Gradient Descent-driven All-in-One Deep Unfolding Networks (CVPR 2025 · similarity 0.85)
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.
