Research map / Image generation / Image restoration
WaveC2R: Wavelet-Driven Coarse-to-Refined Hierarchical Learning for Radar Retrieval
Chunlei Shi, Han Xu, Yinghao Li, Yi-Lin Wei, Yongchao Feng, Yecheng Zhang, Dan Niu
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 |
|---|---|---|---|
| remove WTF module (ablation) | WaveC2R · SEVIR | CSI: 0.311 -> 0.388 | Improved |
| add WTF module (ablation) | WaveC2R · SEVIR | LPIPS: 0.410 -> 0.219 | Improved |
| add VIS channel (ablation) | WTF model · SEVIR | CSI: 0.344 -> 0.360 | Improved |
| add detail-enhanced diffusion refinement (DEDR) (ablation) | WTF + VIS · SEVIR | CSI: 0.360 -> 0.370 | Improved |
| add high-low frequency conditioning (HLF) (ablation) | WTF + VIS + DEDR · SEVIR | CSI: 0.370 -> 0.388 | Improved |
| add HLF to diffusion refinement, SSIM (ablation) | WTF + VIS + DEDR · SEVIR | SSIM: 0.611 -> 0.615 | No clear effect |
| replace MSE with FIBL loss (ablation) | Earthformer · SEVIR (threshold >=219) | CSI: 0.069 -> 0.149 | Improved |
| replace MSE with FIBL loss (ablation) | Smaat-Unet · SEVIR (threshold >=219) | CSI: 0.055 -> 0.144 | Improved |
| replace MSE with FIBL loss (ablation) | AA-TransUnet · SEVIR (threshold >=74) | CSI: 0.506 -> 0.503 | No clear effect |
| replace FACL with FIBL loss (ablation) | Earthformer · SEVIR (Earthformer) | LPIPS: 0.426 -> 0.278 | Improved |
| use WaveC2R vs best deterministic baseline AA-TransUnet MSE | radar retrieval · SEVIR satellite-to-radar retrieval | Avg. CSI: 0.302 -> 0.327 | Improved |
| use GAN-based generators (UVCGAN, Pix2pix) instead of deterministic models | radar retrieval · SEVIR (threshold >=219) | CSI: 0.095 -> 0.015 (UVCGAN), 0.014 (Pix2pix) | Hurt |
Closest papers by meaning
- PrecipDiff: Leveraging Image Diffusion Models to Enhance Satellite-Based Precipitation Observations (AAAI 2025 · similarity 0.85)
- AlphaPre: Amplitude-Phase Disentanglement Model for Precipitation Nowcasting (CVPR 2025 · similarity 0.84)
- Integrating Multi-Source Data for Long Sequence Precipitation Forecasting (AAAI 2025 · similarity 0.84)
- Fourier Amplitude and Correlation Loss: Beyond Using L2 Loss for Skillful Precipitation Nowcasting (NeurIPS 2024 · similarity 0.83)
- Data-driven Precipitation Nowcasting Using Satellite Imagery (AAAI 2025 · similarity 0.83)
- DiffCast: A Unified Framework via Residual Diffusion for Precipitation Nowcasting (CVPR 2024 · similarity 0.83)
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.
