Image Classification on ImageNet-1K 1.0 (val) (Top-1 Accuracy and Hardware Throughput)
82.2Top-1 AccFocal-Tiny
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Focal-TinyParams (M)=29, FLOPs (G)=4.9, Batch size=64, Image resolution=224 x 2242022.05 | 82.2 | 471 | 372 | 261 | 384 | |
| ConvNext-TiParams (M)=28, FLOPs (G)=4.5, Batch size=64, Image resolution=224 x 2242022.05 | 82.1 | 1,654 | 762 | 571 | 1,079 | |
| PVTv2-B2Params (M)=25, FLOPs (G)=4, Batch size=64, Image resolution=224 x 2242022.05 | 82 | 1,175 | 670 | 451 | 854 | |
| XCiT-S12Params (M)=26, FLOPs (G)=4.8, Batch size=64, Image resolution=224 x 2242022.05 | 82 | 1,727 | 761 | 504 | 1,068 | |
| LITv2-SParams (M)=28, FLOPs (G)=3.7, Batch size=64, Image resolution=224 x 2242022.05 | 82 | 1,874 | 1,304 | 928 | 1,471 | |
| CaiT-XS24Params (M)=27, FLOPs (G)=5.4, Batch size=64, Image resolution=224 x 2242022.05 | 81.8 | 991 | 484 | 299 | 623 | |
| CvT-13Params (M)=20, FLOPs (G)=4.5, Batch size=64, Image resolution=224 x 2242022.05 | 81.6 | 1,595 | 716 | 379 | 947 | |
| CoAtNet-0Params (M)=25, FLOPs (G)=4.2, Batch size=64, Image resolution=224 x 2242022.05 | 81.6 | 1,538 | 962 | 643 | 1,151 | |
| Swin-TiParams (M)=28, FLOPs (G)=4.5, Batch size=64, Image resolution=224 x 2242022.05 | 81.3 | 1,564 | 1,039 | 710 | 961 | |
| TNT-SParams (M)=24, FLOPs (G)=5.2, Batch size=64, Image resolution=224 x 2242022.05 | 81.3 | 802 | 431 | 298 | 534 | |
| Twins-PCPVT-SParams (M)=24, FLOPs (G)=3.8, Batch size=64, Image resolution=224 x 2242022.05 | 81.2 | 1,455 | 792 | 529 | 998 | |
| ResNet-50Params (M)=26, FLOPs (G)=4.1, Batch size=64, Image resolution=224 x 2242022.05 | 80.4 | 1,424 | 1,123 | 877 | 1,279 | |
| PVT-SParams (M)=25, FLOPs (G)=3.8, Batch size=64, Image resolution=224 x 2242022.05 | 79.8 | 1,460 | 798 | 548 | 1,007 |