Research map / Probabilistic and scientific modeling / Variational and generative sampling
Taming Flow Matching with Unbalanced Optimal Transport into Fast Pansharpening
Zihan Cao, Yu Zhong, Liang-Jian Deng
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 |
|---|---|---|---|
| train with UOT loss instead of plain flow matching to get one-step generation (ablation) | OTFM · WV3 | SAM: 2.83 -> 2.76 | Improved |
| add pansharpening-regularized cost to UOT (ablation) | OTFM · WV3 full-resolution | HQNR: 0.940 -> 0.954 | Improved |
| use UOT instead of Reflow distillation for one-step sampling (ablation) | OTFM · WV3 | SAM: 3.18 -> 2.76 | Improved |
| use UOT instead of consistency model for one-step sampling (ablation) | OTFM · WV3 | SAM: 2.88 -> 2.76 | Improved |
| use UOT instead of NOT with flow matching (ablation) | OTFM · WV3 | SAM: 2.89 -> 2.76 | Improved |
| use UOT instead of GAN discriminator with flow matching (ablation) | OTFM · WV3 | SAM: 2.85 -> 2.76 | Improved |
| replace flow matching with diffusion in same architecture, without UOT (ablation) | OTFM · WV3 | SAM: 2.87 vs 2.76 | Hurt |
| use one-step OTFM instead of 25-step DDIF | pansharpening · WV3 reduced-resolution | SAM: 2.74 -> 2.76 | No clear effect |
| use one-step OTFM instead of PanDiff (1000 steps) | pansharpening · 256x256 image | latency (s): 261.410 -> 0.029 | Improved |
Closest papers by meaning
- SSDiff: Spatial-spectral Integrated Diffusion Model for Remote Sensing Pansharpening (NeurIPS 2024 · similarity 0.90)
- Dual-Granularity Semantic Guided Sparse Routing Diffusion Model for General Pansharpening (CVPR 2025 · similarity 0.87)
- Hyperspectral Pansharpening via Diffusion Models with Iteratively Zero-Shot Guidance (CVPR 2025 · similarity 0.86)
- Building Bridges across Spatial and Temporal Resolutions: Reference-Based Super-Resolution via Change Priors and Conditional Diffusion Model (CVPR 2024 · similarity 0.86)
- Adaptive Flow Matching for Resolving Small-Scale Physics (ICML 2025 · similarity 0.85)
- SCoT: Unifying Consistency Models and Rectified Flows via Straight-Consistent Trajectories (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.
