Image Classification on ImageNet 1K (Latency, Power, and Throughput Efficiency)
77.4AccuracySHViT-S3
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| SHViT-S3Type=Hybrid, Inp. Res.=224, Param. (M)=14.1, FLOPs (M)=3662026.05 | 77.4 | 7.35 | 1,245.9 | 29 | 22.6 | 3.42 | |
| MicroViTv2-S3Type=Hybrid, Inp. Res.=224, Param. (M)=17.0, FLOPs (M)=6762026.05 | 77.4 | 5.9 | 1,335.5 | 29.9 | 22.5 | 3.44 | |
| MicroViTv1-S3Type=Hybrid, Inp. Res.=224, Param. (M)=16.7, FLOPs (M)=5802026.05 | 77.1 | 5.9 | 1,326.2 | 29.8 | 22.5 | 3.43 | |
| FastViT-T8Type=Hybrid, Inp. Res.=256, Param. (M)=4.0, FLOPs (M)=6872026.05 | 76.7 | 2.67 | 665.6 | 29.8 | 44.8 | 1.7 | |
| Fasternet-T1Type=CNN, Inp. Res.=224, Param. (M)=7.6, FLOPs (M)=8512026.05 | 76.2 | 4.41 | 1,314.8 | 31.4 | 23.9 | 3.19 | |
| SigLIPZero-shot=true2025.09 | 76 | — | — | — | — | — | |
| MobileNetV3-LType=CNN, Inp. Res.=224, Param. (M)=3.5, FLOPs (M)=3142026.05 | 75.2 | 3.49 | 941.5 | 26.7 | 28.4 | 2.65 | |
| SHViT-S2Type=Hybrid, Inp. Res.=224, Param. (M)=11.5, FLOPs (M)=3662026.05 | 75.2 | 6.56 | 1,666.1 | 25.3 | 15.2 | 4.95 | |
| RepVGG-B0Type=CNN, Inp. Res.=224, Param. (M)=14.3, FLOPs (M)=3.1 G2026.05 | 75.1 | 4.2 | 997.1 | 33.9 | 33.9 | 2.21 | |
| MicroViTv2-S2Type=Hybrid, Inp. Res.=224, Param. (M)=12.7, FLOPs (M)=4072026.05 | 75.1 | 5.25 | 1,883.3 | 28 | 14.9 | 5.04 | |
| EdgeNeXt-XSType=Hybrid, Inp. Res.=256, Param. (M)=2.3, FLOPs (M)=5362026.05 | 75 | 5.96 | 573.9 | 27.6 | 48.2 | 1.56 | |
| MobileViT-XSType=Hybrid, Inp. Res.=256, Param. (M)=2.3, FLOPs (M)=9352026.05 | 74.8 | 4.75 | 376.1 | 30.7 | 26.8 | 2.79 | |
| MicroViTv1-S2Type=Hybrid, Inp. Res.=224, Param. (M)=10.0, FLOPs (M)=3452026.05 | 74.6 | 5.25 | 1,843.2 | 27.6 | 15 | 4.97 | |
| iFormer-TType=Hybrid, Inp. Res.=224, Param. (M)=2.9, FLOPs (M)=5302026.05 | 74.3 | 4.12 | 1,016.7 | 29.8 | 29.4 | 2.53 | |
| EfficientViT-M4Type=Hybrid, Inp. Res.=224, Param. (M)=8.8, FLOPs (M)=3032026.05 | 74.3 | 9.17 | 1,687.1 | 25.1 | 15 | 4.95 | |
| EfficientFormerV2-S0Type=Hybrid, Inp. Res.=224, Param. (M)=3.6, FLOPs (M)=4072026.05 | 73.7 | 4.64 | 900.7 | 30.8 | 26.4 | 2.79 | |
| SHViT-S1Type=Hybrid, Inp. Res.=224, Param. (M)=6.3, FLOPs (M)=2412026.05 | 72.8 | 6.41 | 2,110.9 | 24.2 | 11.5 | 6.33 | |
| MicroViTv2-S1Type=Hybrid, Inp. Res.=224, Param. (M)=6.7, FLOPs (M)=2502026.05 | 72.7 | 4.84 | 2,367.6 | 25.8 | 10.8 | 6.67 | |
| MicroViTv1-S1Type=Hybrid, Inp. Res.=224, Param. (M)=6.4, FLOPs (M)=2312026.05 | 72.6 | 4.84 | 2,366.9 | 26 | 11 | 6.6 | |
| RepVGG-A0Type=CNN, Inp. Res.=224, Param. (M)=8.3, FLOPs (M)=1.4 G2026.05 | 72.4 | 3.1 | 1,562.9 | 31.9 | 20 | 3.62 | |
| MobileNetV2-1.0Type=CNN, Inp. Res.=224, Param. (M)=3.5, FLOPs (M)=3142026.05 | 72 | 2.58 | 781.7 | 27.1 | 34.6 | 2.08 | |
| Fasternet-T0Type=CNN, Inp. Res.=224, Param. (M)=3.9, FLOPs (M)=3402026.05 | 71.9 | 4 | 1,955.2 | 28.1 | 14.4 | 4.99 | |
| EdgeNeXt-XXSType=Hybrid, Inp. Res.=256, Param. (M)=1.3, FLOPs (M)=2612026.05 | 71.2 | 4.65 | 865.2 | 25.6 | 29.6 | 2.41 | |
| MobileViTV2-0.5Type=Hybrid, Inp. Res.=256, Param. (M)=1.4, FLOPs (M)=4802026.05 | 70.2 | 5.29 | 666.3 | 28.4 | 42.6 | 1.65 | |
| MobileViT-XXSType=Hybrid, Inp. Res.=256, Param. (M)=1.3, FLOPs (M)=2612026.05 | 69 | 4.34 | 699.5 | 28.2 | 40.3 | 1.71 | |
| LongCLIPZero-shot=true2025.09 | 66.9 | — | — | — | — | — | |
| CLIPZero-shot=true2025.09 | 63.1 | — | — | — | — | — | |
| NegCLIPZero-shot=true2025.09 | 61 | — | — | — | — | — | |
| LSSZero-shot=true2025.09 | 60.8 | — | — | — | — | — | |
| DreamLIPZero-shot=true2025.09 | 55.7 | — | — | — | — | — | |
| DCIP1Zero-shot=true2025.09 | 53.3 | — | — | — | — | — | |
| DACLLMZero-shot=true2025.09 | 51.1 | — | — | — | — | — | |
| CE-CLIPZero-shot=true2025.09 | 50 | — | — | — | — | — |