Research map / Image generation / Image restoration
Rubik's Cube: High-Order Channel Interactions with a Hierarchical Receptive Field
Naishan Zheng, Man Zhou, Chong Zhou, Chen Change Loy
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 |
|---|---|---|---|
| replace standard convolution with Rubik's cube convolution | DRBN · LOL | PSNR: 19.7931 -> 20.3769 | Improved |
| replace standard convolution with Rubik's cube convolution | SID · LOL | PSNR: 20.1062 -> 20.4972 | Improved |
| replace standard convolution with Rubik's cube convolution | DRBN · LOL -> Huawei generalization | PSNR: 17.7929 -> 17.9811 | Improved |
| use RubikConv instead of parameter-matched Conv1x1 variant (ablation) | DRBN · LOL | PSNR: 19.8648 -> 20.3769 | Improved |
| use RubikConv instead of parameter-matched Conv1x1 variant (ablation) | SID · LOL | PSNR: 19.9709 -> 20.4972 | Improved |
| replace Conv1x1 four-layer stand-in with original convolution (ablation) | DRBN · LOL -> Huawei | PSNR: 17.7929 -> 16.2748 | Hurt |
| replace standard convolution with Rubik's cube convolution | Restormer · SIDD | PSNR: 40.0124 -> 40.0853 | No clear effect |
| replace standard convolution with Rubik's cube convolution | DnCNN · SIDD | PSNR: 37.1992 -> 37.4123 | Improved |
| replace standard convolution with Rubik's cube convolution | PanNet · GaoFen2 | PSNR: 42.1699 -> 43.1764 | Improved |
| replace standard convolution with Rubik's cube convolution | INNFormer · WorldView-II | PSNR: 41.5363 -> 41.5720 | No clear effect |
| replace standard convolution with Rubik's cube convolution | MPRNet · GoPro | PSNR: 32.6546 -> 32.7197 | Improved |
| increase shifted pixels p from 1 to 3 (ablation) | RubikConv on SID · LOL | PSNR: 20.4972 -> 20.4858 | No clear effect |
Closest papers by meaning
- xUnit: Learning a Spatial Activation Function for Efficient Image Restoration (CVPR 2018 · similarity 0.87)
- Toward DNN of LUTs: Learning Efficient Image Restoration With Multiple Look-Up Tables (TPAMI 2024 · similarity 0.87)
- Correlation Matching Transformation Transformers for UHD Image Restoration (AAAI 2024 · similarity 0.86)
- Sharing Key Semantics in Transformer Makes Efficient Image Restoration (NeurIPS 2024 · similarity 0.85)
- Omni-Kernel Network for Image Restoration (AAAI 2024 · similarity 0.85)
- TinyLUT: Tiny Look-Up Table for Efficient Image Restoration at the Edge (NeurIPS 2024 · 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.
