Research map / Image generation / Super-resolution
ProSR: Semantic-Prototype-Guided Discrete Modeling for Physically Consistent SAR Super-Resolution
Byoungwoo Kim, Munchurl Kim
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 Detail-VQ instead of standard HR-VQ (ablation) | ProSR AE · Stage 1 AE reconstruction | PSNR: 16.5177 -> 18.1018 | Improved |
| use Detail-VQ instead of Dual-VQ (ablation) | ProSR AE · Stage 1 AE reconstruction | SSIM: 0.3296 -> 0.3799 | Improved |
| increase codebook size from 512 to 1024 (ablation) | ProSR AE · Stage 1 AE reconstruction | PSNR: 17.9659 -> 18.1018 | Improved |
| increase codebook size from 1024 to 2048 (ablation) | ProSR AE · Stage 1 AE reconstruction | PSNR: 18.1018 -> 18.0894 | No clear effect |
| add PMGA instead of vanilla cross-attention (ablation) | ProSR · SAR ISR x4 benchmark | FID: 25.39 -> 23.70 | Improved |
| add PMGA instead of vanilla cross-attention (ablation) | ProSR · SAR ISR x4 benchmark | Dens: 0.9404 -> 0.9741 | Improved |
| add PMGA instead of vanilla cross-attention (ablation) | ProSR · SAR ISR x4 benchmark | IW-SSIM: 0.4186 -> 0.4206 | Improved |
| use discrete token prediction instead of continuous DiT-style modeling (ablation) | ProSR · SAR ISR x4 benchmark | FID: 40.43 -> 23.70 | Improved |
| add stochastic LR-patch training in the AE (p=0.25 vs p=0) (ablation) | ProSR AE · Stage 1 AE reconstruction | SSIM: 0.1024 -> 0.3799 | Improved |
| raise stochastic training probability from 0.25 to 0.5 (ablation) | ProSR AE · Stage 1 AE reconstruction | PSNR/LPIPS: PSNR 18.1018 -> 18.1822, LPIPS 0.2044 -> 0.2159 | Mixed |
| use ProSR outputs instead of strongest baseline UPSR for ATR | ProSR · MSTAR ATR transfer | accuracy (%): 84.18 -> 88.39 | Improved |
| use prototype token count K=3 instead of K=2 or K>=4 (ablation) | SPMG · semantic prototype maps | semantic map quality | Improved |
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.
- ResShift — ResShift: Efficient Diffusion Model for Image Super-resolution by Residual Shifting (NeurIPS 2023) · better on 3 of 3 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
- FUSAR-GPT: A Spatiotemporal Feature-Embedded and Two-Stage Decoupled Visual Language Model for SAR Imagery (CVPR 2026 · similarity 0.85)
- Visual Autoregressive Modeling for Image Super-Resolution (ICML 2025 · similarity 0.83)
- SARDet-100K: Towards Open-Source Benchmark and ToolKit for Large-Scale SAR Object Detection (NeurIPS 2024 · similarity 0.82)
- RSAR: Restricted State Angle Resolver and Rotated SAR Benchmark (CVPR 2025 · similarity 0.82)
- Building Bridges across Spatial and Temporal Resolutions: Reference-Based Super-Resolution via Change Priors and Conditional Diffusion Model (CVPR 2024 · similarity 0.81)
- Cross-Rejective Open-Set SAR Image Registration (CVPR 2025 · similarity 0.81)
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.
