Research map / Vision-language and multimodal / Cross-modal learning
ReCoFuse: Ultra-Robust Image Fusion via Restorative Multi-Modal Diffusion Reciprocal Coupling
Hao Zhang, Shuhan Yang, Linfeng Tang, Xunpeng Yi, Jiayi Ma
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 TIM cross-modal coupling to the diffusion restoration branches (ablation) | ReCoFuse visible restoration · MFNet | PSNR: 24.512 -> 25.205 | Improved |
| add TIM cross-modal coupling to the diffusion restoration branches (ablation) | ReCoFuse infrared restoration · MFNet | FID: 39.190 -> 25.074 | Improved |
| insert TIM at each sampling step instead of late fusion after sampling (ablation) | ReCoFuse · MFNet fusion | MI: MI 3.083 -> 3.114 | Improved |
| add time embedding to TIM (ablation) | ReCoFuse · MFNet fusion | MI: 2.464 -> 3.114 | Improved |
| alternating regularization instead of separate-manner optimization (ablation) | ReCoFuse · MFNet fusion | MI: 2.553 -> 3.114 | Improved |
| alternating regularization instead of joint optimization (ablation) | ReCoFuse · MFNet fusion | MI: 2.972 -> 3.114 | Improved |
| alternating regularization instead of alternately freezing one module (decoupling) (ablation) | ReCoFuse · MFNet fusion | MI: 2.000 -> 3.114 | Improved |
| use contrast loss instead of proportional intensity loss (ablation) | ReCoFuse · MFNet fusion | MI: 2.436 -> 3.114 | Improved |
| use texture loss instead of proportional gradient loss (ablation) | ReCoFuse · MFNet fusion | MI: 2.458 -> 3.114 | Improved |
| use ReCoFuse fused images for detection | YOLO-v5 on LLVIP · LLVIP detection | mAP@.75: 0.715 -> 0.718 (IR only vs ReCoFuse) | Improved |
| use ReCoFuse fused images for segmentation | SegNeXt on MFNet · MFNet segmentation | mIoU: 54.94 -> 57.67 (VIS only vs ReCoFuse) | Improved |
| ReCoFuse versus baselines on real-world AWMM-100k | ReCoFuse · AWMM-100k | MI: 3.095 vs Diff-IF 3.826 | Mixed |
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.
- MRFS — MRFS: Mutually Reinforcing Image Fusion and Segmentation (CVPR 2024) · 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
- MODULE: A Mutual-Promoting Deep Unfolding Framework Towards Degradation-Robust Multi-modal Image Fusion (NeurIPS 2026 · similarity 0.89)
- Degradation-Robust Fusion: An Efficient Degradation-Aware Diffusion Framework for Multimodal Image Fusion in Arbitrary Degradation Scenarios (CVPR 2026 · similarity 0.87)
- UniLDiff: Unlocking the Power of Diffusion Priors for All-in-One Image Restoration (CVPR 2026 · similarity 0.87)
- A Modular Conditional Diffusion Framework for Image Reconstruction (NeurIPS 2024 · similarity 0.86)
- Efficient Rectified Flow for Image Fusion (NeurIPS 2025 · similarity 0.85)
- Restabilizing Diffusion Models with Predictive Noise Fusion Strategy for Image Super-Resolution (AAAI 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.
