Research map / Image generation / Super-resolution
Exploring Sparsity in Image Super-Resolution for Efficient Inference
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 jointly learned spatial and channel masks to skip redundant computation | SMSR · Set14, ×2 | PSNR: 33.65 -> 33.64 | Helped |
| remove spatial and channel masks (ablation) | SMSR · Set14, ×2 | FLOPs: 1.00× -> 0.61× | Hurt |
| add channel masks without spatial masks (ablation) | SMSR · Set14, ×2 | PSNR: 33.65 -> 33.53 | Hurt |
| add spatial masks without channel masks (ablation) | SMSR · Set14, ×2 | PSNR: 33.65 -> 33.60 | Hurt |
| increase sparsity regularization weight λ₀ from 0.1 to 0.2 (ablation) | SMSR · Set14, ×2 | PSNR: 33.64 -> 33.61; 0.46 -> 0.64 sparsity | Mixed |
| increase sparsity regularization weight λ₀ from 0.1 to 0.3 (ablation) | SMSR · Set14, ×2 | PSNR: 33.64 -> 33.52; 0.46 -> 0.73 sparsity | Mixed |
| replace learned spatial masks with fixed gradient-based masks (ablation) | SMSR · Set14, ×2 | PSNR: 33.64 vs. 33.33/33.30 | Hurt |
| use SMSR rather than FALSR-A | lightweight SR models · Set14, ×2 | PSNR: 33.55 -> 33.64 | Helped |
| use SMSR rather than CARN | lightweight SR models · Set14, ×4 | PSNR: 28.60 -> 28.55 | Hurt |
| use SMSR rather than CARN | lightweight SR models · Set14, ×3 | PSNR: 30.29 -> 30.33 | Helped |
Closest papers by meaning
- Aligned Structured Sparsity Learning for Efficient Image Super-Resolution (NeurIPS 2021 · similarity 0.89)
- Lightweight Image Super-Resolution via Flexible Meta Pruning (ICML 2024 · similarity 0.87)
- Learning a Single Convolutional Super-Resolution Network for Multiple Degradations (CVPR 2018 · similarity 0.87)
- Learning Efficient Image Super-Resolution Networks via Structure-Regularized Pruning (ICLR 2022 · similarity 0.87)
- AdaFormer: Efficient Transformer with Adaptive Token Sparsification for Image Super-resolution (AAAI 2024 · similarity 0.87)
- Image Processing GNN: Breaking Rigidity in Super-Resolution (CVPR 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.