Research map / Image generation / Image restoration
V-DESIRR: Very Fast Deep Embedded Single Image Reflection Removal
B H Pawan Prasad, Green Rosh K S, Lokesh R. Boregowda, Kaushik Mitra, Sanjoy Chowdhury
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 HSSNet higher-scale stage on top of LSSNet (ablation) | LSSNet · Real datasets average | PSNR: 24.789 -> 25.069 | Improved |
| use LSSNet instead of HSSNet at higher scales (no CGF) (ablation) | LSSNet + HSSNet · Real datasets average | PSNR: 25.069 -> 25.412 | Improved |
| add CGF upsampling to LSSNet + HSSNet (ablation) | LSSNet + HSSNet · Real datasets average | PSNR: 25.069 -> 26.323 | Improved |
| add CGF upsampling to LSSNet + LSSNet (ablation) | LSSNet + LSSNet · Real datasets average | PSNR: 25.412 -> 26.45 | Improved |
| use shallow HSSNet instead of LSSNet at higher scales (with CGF) (ablation) | LSSNet + CGF · Real datasets average | PSNR: 26.45 -> 26.323; 2.5x faster | Mixed |
| add CGF upsampling (ablation) | LSSNet + HSSNet · Real datasets average | SSIM: 0.897 -> 0.898 | No clear effect |
| add HSSNet higher-scale stage on top of LSSNet (ablation) | LSSNet · Real datasets average | SSIM: 0.878 -> 0.897 | Improved |
| increase number of scales N for 64MP input (ablation) | V-DESIRR · 64MP images | visual quality | Improved |
| use proposed scale-space method instead of RAGNet | RAGNet · Average of four datasets | PSNR: 24.79 -> 26.45 | Improved |
| use proposed method instead of RAGNet | RAGNet · 540x400 tile | processing time (s): 0.331 -> 0.0149 | Improved |
| use proposed method instead of RAGNet on Real20 | RAGNet · Real20 | PSNR: 21.26 -> 25.06 | Improved |
| quantize model with enhanced quantizer for smartphone | V-DESIRR · 12MP images | visual quality | No clear effect |
Closest papers by meaning
- LightRR: A Lightweight Network for Single Image Reflection Removal (CVPR 2026 · similarity 0.89)
- A Generic Deep Architecture for Single Image Reflection Removal and Image Smoothing (ICCV 2017 · similarity 0.89)
- Seeing Deeply and Bidirectionally: A Deep Learning Approach for Single Image Reflection Removal (ECCV 2018 · similarity 0.89)
- A Lightweight Deep Exclusion Unfolding Network for Single Image Reflection Removal (TPAMI 2025 · similarity 0.88)
- Revisiting Single Image Reflection Removal In the Wild (CVPR 2024 · similarity 0.86)
- TinyLUT: Tiny Look-Up Table for Efficient Image Restoration at the Edge (NeurIPS 2024 · similarity 0.86)
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.
