Research map / Image generation / Image restoration
LightRR: A Lightweight Network for Single Image Reflection Removal
Wenbin Yin, Junkang Zhang, Sunzhe Yang, Faming Fang, Guixu Zhang
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 asymmetric frequency SSM (Mamba on LL band only) instead of full-resolution 2D-SSM (ablation) | LightRR · Average (with Nature) | PSNR: 25.95 -> 26.39 | Improved |
| use AF-SSM instead of full-resolution 2D-SSM (ablation) | LightRR · Average (with Nature) | peak GPU memory (MB): 361.03 -> 170.72 | Improved |
| use AF-SSM instead of full-resolution 2D-SSM (ablation) | LightRR · Average (with Nature) | runtime (ms): 51.37 -> 74.57 | Hurt |
| replace sub-pixel conv sampling with wavelet WDD/WSU and dual-frequency skips (ablation) | LightRR · Average (with Nature) | PSNR: 26.17 -> 26.39 | Improved |
| replace sub-pixel conv sampling with wavelet WDD/WSU (ablation) | LightRR · Average (with Nature) | GFLOPs: 14.47 -> 9.62 | Improved |
| add knowledge distillation from VGG19 versus no pretrained features (ablation) | LightRR · Average (with Nature) | PSNR: 25.57 -> 26.39 | Improved |
| distill VGG19 features instead of directly injecting them (ablation) | LightRR · Average (with Nature) | PSNR: 25.98 -> 26.39 | Improved |
| directly inject pretrained VGG19 features versus none (ablation) | LightRR · Average (with Nature) | PSNR: 25.57 -> 25.98 | Improved |
| use LightRR instead of scaled-down DSIT-light at similar parameter budget | LightRR · Average (with Nature) | PSNR: 25.16 -> 26.39 | Improved |
Closest papers by meaning
- A Lightweight Deep Exclusion Unfolding Network for Single Image Reflection Removal (TPAMI 2025 · similarity 0.89)
- V-DESIRR: Very Fast Deep Embedded Single Image Reflection Removal (ICCV 2021 · similarity 0.89)
- Single Image Reflection Removal Exploiting Misaligned Training Data and Network Enhancements (CVPR 2019 · similarity 0.88)
- Revisiting Single Image Reflection Removal In the Wild (CVPR 2024 · similarity 0.88)
- Single Image Reflection Removal Beyond Linearity (CVPR 2019 · similarity 0.88)
- Efficient Visual State Space Model for Image Deblurring (CVPR 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.
