Image Classification on ImageNet-1K (val) (Top-1 Accuracy and Hardware Throughput)
0.854Top-1 AccuracyFasterViT-4
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| FasterViT-4Resolution=224, Params=425M, MACs=36.6G2024.11 | 0.854 | 50 | 800 | 1,392 | 18 | 234 | |
| MOAT-3Resolution=224, Params=190M, MACs=44.9G2024.11 | 0.853 | 70 | 805 | 962 | 21 | 246 | |
| AsCAN-LResolution=224, Params=173M, MACs=30.7G2024.11 | 0.8524 | 120 | 1,381 | 1,617 | 40 | 440 | |
| MaxViT-LResolution=224, Params=212M, MACs=43.9G2024.11 | 0.8517 | 34 | 544 | 759 | 10 | 123 | |
| EfficientNetV2-MResolution=480, Params=55M, MACs=24.0G2024.11 | 0.851 | 84 | 918 | 1,094 | 26 | 329 | |
| MogaNet-XLResolution=224, Params=181M, MACs=34.5G2024.11 | 0.851 | 32 | 471 | 576 | 19 | 210 | |
| MaxViT-BResolution=224, Params=120M, MACs=23.4G2024.11 | 0.8495 | 34 | 507 | 1,012 | 11 | 164 | |
| FasterViT-3Resolution=224, Params=160M, MACs=18.2G2024.11 | 0.849 | 46 | 831 | 3,131 | 17 | 257 | |
| RDNet-LResolution=224, Params=186M, MACs=34.7G2024.11 | 0.848 | 78 | 640 | 990 | 32 | 290 | |
| AsCAN-BResolution=224, Params=98M, MACs=16.7G2024.11 | 0.8473 | 113 | 1,878 | 2,393 | 38 | 590 | |
| NFNet-F1Resolution=320, Params=132M, MACs=35.5G2024.11 | 0.847 | 106 | 998 | 1,192 | 26 | 48 | |
| MOAT-2Resolution=224, Params=73M, MACs=17.2G2024.11 | 0.847 | 132 | 1,367 | 2,087 | 38 | 424 | |
| MogaNet-LResolution=224, Params=83M, MACs=15.9G2024.11 | 0.847 | 34 | 523 | 882 | 24 | 290 | |
| Swin-BResolution=384, Params=88M, MACs=47.0G2024.11 | 0.845 | 51 | 458 | 486 | 20 | 85 | |
| CoAtNet-3Resolution=224, Params=168M, MACs=34.7G2024.11 | 0.845 | 132 | 1,085 | 1,105 | 37 | 388 | |
| MaxViT-SResolution=224, Params=69M, MACs=11.7G2024.11 | 0.8445 | 70 | 1,019 | 1,775 | 24 | 243 | |
| RDNet-BResolution=224, Params=87M, MACs=15.4G2024.11 | 0.844 | 80 | 1,891 | — | 48 | 589 | |
| EfficientNet-B7Resolution=600, Params=66M, MACs=37.0G2024.11 | 0.843 | 72 | 321 | 350 | 24 | 124 | |
| ConvNeXt-LResolution=224, Params=198M, MACs=34.4G2024.11 | 0.843 | 168 | 1,045 | 1,127 | 58 | 362 | |
| SMT-BResolution=224, Params=32M, MACs=7.7G2024.11 | 0.843 | 27 | 388 | 1,412 | 14 | 243 | |
| BiFormer-BResolution=224, Params=57M, MACs=9.8G2024.11 | 0.843 | 50 | 840 | 1,439 | 28 | 440 | |
| FasterViT-2Resolution=224, Params=76M, MACs=8.7G2024.11 | 0.842 | 64 | 1,112 | 4,376 | 24 | 321 | |
| RMT-SResolution=224, Params=27M, MACs=4.5G2024.11 | 0.841 | 46 | 790 | 2,439 | 38 | 480 | |
| CoAtNet-2Resolution=224, Params=75M, MACs=15.7G2024.11 | 0.841 | 133 | 1,718 | 2,040 | 38 | 540 | |
| EfficientNet-B6Resolution=528, Params=43M, MACs=19.0G2024.11 | 0.84 | 88 | 529 | 589 | 27 | 205 | |
| EfficientNetV2-SResolution=384, Params=24M, MACs=8.8G2024.11 | 0.839 | 112 | 1,884 | 2,744 | 34 | 561 | |
| ConvNeXt-BResolution=224, Params=89M, MACs=15.4G2024.11 | 0.838 | 167 | 1,760 | 2,073 | 58 | 619 | |
| PVTv2-B5Resolution=224, Params=82M, MACs=11.8G2024.11 | 0.838 | 32 | 511 | 2,112 | 17 | 291 | |
| BiFormer-SResolution=224, Params=26M, MACs=4.5G2024.11 | 0.838 | 45 | 937 | 2,139 | 36 | 595 | |
| RDNet-SResolution=224, Params=50M, MACs=8.7G2024.11 | 0.837 | 102 | 1,782 | 2,761 | 53 | 780 | |
| SMT-SResolution=224, Params=21M, MACs=4.7G2024.11 | 0.837 | 38 | 566 | 2,211 | 43 | 348 | |
| MaxViT-TResolution=224, Params=31M, MACs=5.6G2024.11 | 0.8362 | 73 | 1,098 | 2,756 | 23 | 357 | |
| NFNet-F0Resolution=256, Params=72M, MACs=12.4G2024.11 | 0.836 | 199 | 2,470 | 3,348 | 47 | 813 | |
| EfficientViT-B3Resolution=224, Params=49M, MACs=4G2024.11 | 0.835 | 114 | 1,882 | 2,860 | 32 | 552 | |
| AsCAN-TResolution=224, Params=55M, MACs=7.7G2024.11 | 0.8344 | 199 | 3,224 | 4,295 | 67 | 1,148 | |
| CoAtNet-1Resolution=224, Params=42M, MACs=8.4G2024.11 | 0.833 | 141 | 2,221 | 2,907 | 45 | 629 | |
| PVTv2-B3Resolution=224, Params=45M, MACs=7G2024.11 | 0.832 | 49 | 933 | 3,782 | 33 | 548 | |
| FasterViT-1Resolution=224, Params=53M, MACs=5.3G2024.11 | 0.832 | 67 | 1,123 | 4,106 | 23 | 363 | |
| ConvNeXt-SResolution=224, Params=50M, MACs=8.7G2024.11 | 0.831 | 174 | 2,291 | 2,889 | 56 | 836 | |
| DeiT-BResolution=384, Params=86M, MACs=55.4G2024.11 | 0.831 | 189 | 1,058 | 1,192 | 130 | 488 | |
| Swin-SResolution=224, Params=50M, MACs=8.7G2024.11 | 0.83 | 50 | 841 | 2,221 | 34 | 436 | |
| EfficientViT-B2Resolution=224, Params=24M, MACs=1.6G2024.11 | 0.821 | 142 | 2,171 | 5,132 | 48 | 782 | |
| CoAtNet-0Resolution=224, Params=25M, MACs=4.2G2024.11 | 0.816 | 214 | 3,537 | 5,221 | 61 | 976 | |
| ViT-B/16Resolution=384, Params=86M, MACs=55.4G2024.11 | 0.779 | 212 | 1,006 | 1,266 | 104 | 86 | |
| ViT-B/32Resolution=384, Params=307M, MACs=190.7G2024.11 | 0.765 | 112 | 893 | 924 | 54 | 27 |