Table 9 Comparison of model complexity on B100 for scale factor \(\times \)3.
From: Lightweight interactive feature inference network for single-image super-resolution
Type | Model | Params | Multi-adds | PSNR(dB)/SSIM | Time (s) |
|---|---|---|---|---|---|
CNN-based | CARN | 1592K | 118.8G | 29.06/0.8034 | 0.4623 |
AWSRN-S | 477K | 48.6G | 28.92/0.8009 | 0.8807 | |
AWSRN-M | 1143K | 116.6G | 29.13/0.8059 | 1.7432 | |
OISR-RK2-s | 1557 K | 160.1 G | 29.10/0.8053 | 0.5304 | |
A2F-S | 324k | 32.3G | 28.92/0.8006 | 0.6651 | |
LESRCNN | 516K | 49.1G | 28.91/0.8005 | 0.2911 | |
LMAN-s | 709K | 73.5G | 29.02/0.8030 | 1.6537 | |
Cross-SRN | 1285K | 130.5G | 29.09/0.8050 | 2.2345 | |
ACNet | 1541K | 369G | 28.98/0.8023 | 0.6436 | |
FMEN | 757K | 77.2G | 29.17/0.8063 | 0.4598 | |
AFAN | 1208K | 143.1G | 29.11/0.8064 | 2.9098 | |
CFGN | 609K | 59.0G | 29.16/0.8066 | 2.1632 | |
Transformer-based | DRSAN-48m | 1292K | 133.4G | 29.18/0.8079 | 25.9941 |
ESRT | 770K | 135G | 29.15/0.8063 | 2.4120 | |
LBNet | 736K | 68.4G | 29.13/0.8061 | 3.2489 | |
Ngswin | 1007K | 66.6G | 29.19/0.8078 | 8.0806 | |
IFIN-S(ours) | 459K | 51.0G | 29.13/0.8064 | 1.6288 | |
IFIN(ours) | 980K | 107.0G | 29.21/0.8082 | 3.1728 |
