Research map / Image generation / Image restoration
Semantics and Content Matter: Towards Multi-Prior Hierarchical Mamba for Image Deraining
Zhaocheng Yu, Kui Jiang, Junjun Jiang, Xianming Liu, Guanglu Sun, Yi Xiao
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 frequency-domain branch (FFCM) to HMM (ablation) | HMM · Rain200H | PSNR: 30.56 -> 33.06 | Improved |
| add grouped DW-Conv and VSSM local-global strategy (ablation) | HMM · Rain200H | PSNR: 32.93 -> 33.06 | Improved |
| use concatenation with 1x1 conv instead of addition for branch fusion (ablation) | HMM · Rain200H | PSNR: 32.82 -> 33.06 | Improved |
| use concatenation instead of cross-attention for branch fusion (ablation) | HMM · Rain200H | PSNR: 32.22 -> 33.06 | Improved |
| add DINOv2 visual prior Pv (ablation) | MPHM · Rain200H | PSNR: 33.06 -> 33.22 | Improved |
| add CLIP textual prior Pt (ablation) | MPHM · Rain200H | PSNR: 33.06 -> 33.20 | Improved |
| inject both Pv and Pt jointly (ablation) | MPHM · Rain200H | PSNR: 33.06 -> 33.53 | Improved |
| use hierarchical sequential prior fusion instead of addition (ablation) | PFI · Rain200H | PSNR: 31.95 -> 33.53 | Improved |
| use hierarchical sequential prior fusion instead of concatenation (ablation) | PFI · Rain200H | PSNR: 32.05 -> 33.53 | Improved |
| use hierarchical two-stage fusion instead of single-stage cross-attention (ablation) | PFI · Rain200H | PSNR: 33.15 -> 33.53 | Improved |
| add multi-prior and HMM design on top of strongest Mamba baselines | MPHM vs FourierMamba · DDN-Data | PSNR: 35.58 vs 34.80 | Hurt |
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.
- FourierMamba — FourierMamba: Fourier Learning Integration with State Space Models for Image Deraining (ICML 2025) · better on 3 of 5 shared results
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
- Harnessing Joint Rain-/Detail-aware Representations to Eliminate Intricate Rains (ICLR 2024 · similarity 0.87)
- Mamba as a Bridge: Where Vision Foundation Models Meet Vision Language Models for Domain-Generalized Semantic Segmentation (CVPR 2025 · similarity 0.86)
- Mamba-Reg: Vision Mamba Also Needs Registers (CVPR 2025 · similarity 0.86)
- Every SAM Drop Counts: Embracing Semantic Priors for Multi-Modality Image Fusion and Beyond (CVPR 2025 · similarity 0.85)
- Rethinking Multi-Scale Representations in Deep Deraining Transformer (AAAI 2024 · similarity 0.85)
- Spatial-Mamba: Effective Visual State Space Models via Structure-Aware State Fusion (ICLR 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.
