Research map / Vision-language and multimodal / Cross-modal learning
UniFusion: A Unified Image Fusion Framework with Robust Representation and Source-Aware Preservation
Xingyuan Li, Songcheng Du, Yang Zou, Haoyuan Xu, Zhiying Jiang, Jinyuan Liu
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 |
|---|---|---|---|
| use frozen DINOv3 encoder instead of a four-layer Transformer encoder (ablation) | UniFusion · M3FD | MI: 3.681 -> 4.268 | Improved |
| add modality adapters instead of simple upsampling (ablation) | UniFusion · M3FD | MI: 3.646 -> 4.268 | Improved |
| add reconstruction-alignment branches (ablation) | UniFusion · M3FD | MI: 3.846 -> 4.268 | Improved |
| add bilevel optimization (ablation) | UniFusion · M3FD | MI: 3.924 -> 4.268 | Improved |
| add bilevel optimization (ablation) | UniFusion · MEFB | MI: 6.374 -> 6.861 | Improved |
| add bilevel optimization (ablation) | UniFusion · MFIF | Qabf: 0.583 -> 0.685 | Improved |
| add reconstruction-alignment branches (ablation) | UniFusion · MEFB | CC: 0.393 -> 0.906 | Improved |
| use UniFusion instead of TC-MoA on infrared-visible fusion Qy | UniFusion · M3FD | Qy: 0.983 -> 0.982 | No clear effect |
| use UniFusion (unified, no task-specific tuning) instead of TC-MoA | UniFusion · M3FD | VIF: 0.870 -> 0.899 | Improved |
| use UniFusion on multi-exposure fusion instead of the best baseline | UniFusion · MEFB | MI: 6.575 -> 6.861 | Improved |
Closest papers by meaning
- MODULE: A Mutual-Promoting Deep Unfolding Framework Towards Degradation-Robust Multi-modal Image Fusion (NeurIPS 2026 · similarity 0.89)
- More Than Meets the Eye: A Unified Image Fusion Framework via Semantic-Pixel Entropy Trade-off for Zero-Shot Generalization (CVPR 2026 · similarity 0.88)
- Every SAM Drop Counts: Embracing Semantic Priors for Multi-Modality Image Fusion and Beyond (CVPR 2025 · similarity 0.87)
- A Unified Solution to Video Fusion: From Multi-Frame Learning to Benchmarking (NeurIPS 2025 · similarity 0.86)
- One Model for ALL: Low-Level Task Interaction Is a Key to Task-Agnostic Image Fusion (CVPR 2025 · similarity 0.85)
- Revisiting Generative Infrared and Visible Image Fusion Based on Human Cognitive Laws (NeurIPS 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.
