notesum.ai
Published at December 3CubeFormer: A Simple yet Effective Baseline for Lightweight Image Super-Resolution
cs.CV
Released Date: December 3, 2024
Authors: Jikai Wang1, Huan Zheng, Jianbing Shen
Aff.: 1University of Macau
| Method | Params | BSD100 | Urban100 | Manga109 |
|---|---|---|---|---|
| (K) | PSNR / SSIM | PSNR / SSIM | PSNR / SSIM | |
| IMDN [18] | 694 | 32.19 / 0.8996 | 32.17 / 0.9283 | 38.88 / 0.9774 |
| RFDN-L [27] | 626 | 32.18 / 0.8996 | 32.24 / 0.9290 | 38.95 / 0.9773 |
| MAFFSRN [35] | 402 | 32.14 / 0.8994 | 31.96 / 0.9268 | - / - |
| LatticeNet [30] | 756 | 32.25 / 0.9005 | 32.43 / 0.9302 | - / - |
| SwinIR [25] | 878 | 32.31 / 0.9012 | 32.76 / 0.9340 | 39.12 / 0.9783 |
| RLFN [21] | 527 | 32.22 / 0.9000 | 32.33 / 0.9299 | - / - |
| GASSL-B+ [43] | 689 | 32.28 / 0.9009 | 32.43 / 0.9310 | 39.10 / 0.9781 |
| OSFFNet [45] | 516 | 32.29 / 0.9012 | 32.67 / 0.9331 | 39.09 / 0.9780 |
| Omni-SR [42] | 772 | 32.36 / 0.9020 | 33.05 / 0.9363 | 39.28 / 0.9784 |
| SRConvNet-L [22] | 885 | 32.28 / 0.9010 | 32.59 / 0.9321 | 39.22 / 0.9779 |
| CubeFormer | 778 | 32.36 / 0.9021 | 33.09 / 0.9368 | 39.39 / 0.9786 |
| IMDN [18] | 703 | 29.09 / 0.8046 | 28.17 / 0.8519 | 33.61 / 0.9445 |
| RFDN-L [27] | 633 | 29.11 / 0.8053 | 28.32 / 0.8547 | 33.78 / 0.9458 |
| MAFFSRN [35] | 807 | 29.13 / 0.8061 | 28.26 / 0.8552 | - / - |
| LatticeNet [30] | 765 | 29.15 / 0.8059 | 28.33 / 0.8538 | - / - |
| SwinIR [25] | 886 | 29.20 / 0.8082 | 28.66 / 0.8624 | 33.98 / 0.9478 |
| ESRT [29] | 770 | 29.15 / 0.8063 | 28.46 / 0.8574 | 33.95 / 0.9455 |
| GASSL-B+ [43] | 691 | 29.19 / 0.8070 | 28.39 / 0.8566 | 34.00 / 0.9467 |
| OSFFNet [45] | 524 | 29.21 / 0.8080 | 28.49 / 0.8595 | 34.00 / 0.9470 |
| Omni-SR [42] | 780 | 29.28 / 0.8094 | 28.84 / 0.8656 | 34.22 / 0.9487 |
| SRConvNet-L [22] | 906 | 29.22 / 0.8081 | 28.56 / 0.8600 | 34.17 / 0.9479 |
| CubeFormer | 787 | 29.29 / 0.8099 | 28.88 / 0.8664 | 34.40 / 0.9490 |
| IMDN [18] | 715 | 27.56 / 0.7353 | 26.04 / 0.7838 | 30.45 / 0.9075 |
| RFDN-L [27] | 643 | 27.58 / 0.7363 | 26.20 / 0.7883 | 30.61 / 0.9096 |
| MAFFSRN [35] | 830 | 27.59 / 0.7370 | 26.16 / 0.7887 | - / - |
| LatticeNet [30] | 777 | 27.62 / 0.7367 | 26.25 / 0.7873 | - / - |
| SwinIR [25] | 897 | 27.69 / 0.7406 | 26.47 / 0.7980 | 30.92 / 0.9151 |
| RLFN [21] | 543 | 27.60 / 0.7364 | 26.17 / 0.7877 | - / - |
| ESRT [29] | 751 | 27.69 / 0.7379 | 26.39 / 0.7962 | 30.75 / 0.9100 |
| GASSL-B+ [43] | 694 | 27.66 / 0.7388 | 26.27 / 0.7914 | 30.92 / 0.9122 |
| OSFFNet [45] | 537 | 27.66 / 0.7393 | 26.36 / 0.7950 | 30.84 / 0.9125 |
| Omni-SR [42] | 792 | 27.71 / 0.7415 | 26.64 / 0.8018 | 31.02 / 0.9151 |
| SRConvNet-L [22] | 902 | 27.69 / 0.7402 | 26.47 / 0.7970 | 30.96 / 0.9139 |
| CubeFormer | 799 | 27.76 / 0.7422 | 26.66 / 0.8026 | 31.39 / 0.9174 |