Research map / Image generation / Text-to-image diffusion
Both Diverse and Realism Matter: Physical Attribute and Style Alignment for Rainy Image Generation
Changfeng Yu, Shiming Chen, Yi Chang, Yibing Song, Luxin Yan
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 style alignment (ablation) | PCGNet · Rain1400 deraining with JORDER-E | PSNR: 30.15 -> 32.56 | Improved |
| add Step-moving training (SMT) (ablation) | PCGNet · Rain1400 deraining with JORDER-E | PSNR: 32.56 -> 33.11 | Improved |
| add SMT to attribute encoder (ablation) | PCGNet · Rain1400 attribute estimation | density accuracy: 70.2% -> 84.7% | Improved |
| add attribute-level consistency loss (ablation) | PCGNet · rain attribute control | attribute accuracy: 0.739 -> 0.895 | Improved |
| add adversarial loss (ablation) | PCGNet · FCRealRain generation | FID: 125.7 -> 108.28 | Improved |
| augment deraining training with JRGR generated data | DDN on Rain1400 · Rain1400 | PSNR: 28.98 -> 27.31 | Hurt |
| augment deraining training with PCGNet data | DDN on Rain1400 · Rain1400 | PSNR: 28.98 -> 31.25 | Improved |
| augment MPRNet training with PCGNet data | MPRNet on Rain1400 · Rain1400 | PSNR: 34.43 -> 34.81 | Improved |
| augment with classical augmentations | DDN / JORDER-E / MPRNet · Rain1400 | PSNR: 28.98 -> 28.72 (DDN) | No clear effect |
| augment IDT transformer deraining with PCGNet data | IDT · Rain1400 | PSNR: 33.39 -> 33.74 | Improved |
| train deraining on PCGNet data for unpaired real rain | DDN · FCRealRain | NIQE: 4.612 -> 3.987 | Improved |
Closest papers by meaning
- Closing the Loop: Joint Rain Generation and Removal via Disentangled Image Translation (CVPR 2021 · similarity 0.90)
- Spatial Attentive Single-Image Deraining With a High Quality Real Rain Dataset (CVPR 2019 · similarity 0.90)
- From Sky to the Ground: A Large-scale Benchmark and Simple Baseline Towards Real Rain Removal (ICCV 2023 · similarity 0.89)
- Not Just Streaks: Towards Ground Truth for Single Image Deraining (ECCV 2022 · similarity 0.89)
- From Rain Generation to Rain Removal (CVPR 2021 · similarity 0.89)
- Harnessing Joint Rain-/Detail-aware Representations to Eliminate Intricate Rains (ICLR 2024 · similarity 0.88)
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.
