Image Classification on ImageNet 1k (v2)
84Top-1 AccuracySwinV2-G
Evaluation Results
| Method | Links | |
|---|---|---|
| SwinV2-Gparam=3.0B, pre-train images=IN-22K-ext-70M, pre-train length (#im)=3.5B, pre-train im size=192x192, pre-train time=<0.5k†, fine-tune im size=640x6402021.11 | 84 | |
| ViT-Gparam=1.8B, pre-train images=JFT-3B, pre-train length (#im)=164B, pre-train im size=224x224, pre-train time=>30k, fine-tune im size=518x5182021.11 | 83.33 | |
| SwinV2-Lparam=197M, pre-train images=IN-22K-14M, pre-train length (#im)=1.3B, pre-train im size=192x192, pre-train time=<20†, fine-tune im size=384x3842021.11 | 78.31 | |
| SwinV2-Bparam=88M, pre-train images=IN-22K-14M, pre-train length (#im)=1.3B, pre-train im size=192x192, pre-train time=<30†, fine-tune im size=384x3842021.11 | 78.08 | |
| SwinV1-Lparam=197M, pre-train images=IN-22K-14M, pre-train length (#im)=1.3B, pre-train im size=224x224, pre-train time=<10†, fine-tune im size=384x3842021.11 | 77.46 | |
| SwinV1-Bparam=88M, pre-train images=IN-22K-14M, pre-train length (#im)=1.3B, pre-train im size=224x224, pre-train time=<30†, fine-tune im size=384x3842021.11 | 76.58 | |
| Ours (DINOv2)Data=INet-22k, Arch.=L/142024.06 | 74.8 | |
| SuperCLIPPretrain=L-12.8B, Zero-shot=true2025.12 | 72.8 | |
| CLIPPretrain=L-12.8B, Zero-shot=true2025.12 | 72 | |
| ResNet-50 (A1)Training Procedure=A1, Number of epochs=600, Training resolution=224 x 224, Training time=110h, Peak memory by GPU (MB)=22,095, Numbers of GPU=42021.10 | 68.7 | |
| ResNet-50 (A2)Training Procedure=A2, Number of epochs=300, Training resolution=224 x 224, Training time=55h, Peak memory by GPU (MB)=22,095, Numbers of GPU=42021.10 | 67.9 | |
| MAEData=INet-1k, Arch.=H/142024.06 | 66.5 | |
| ResNet-50 (A3)Training Procedure=A3, Number of epochs=100, Training resolution=160 x 160, Training time=15h, Peak memory by GPU (MB)=11,390, Numbers of GPU=42021.10 | 66.1 | |
| I-JEPAData=INet-1k, Arch.=H/142024.06 | 66 | |
| SuperCLIPPretrain=L-512M, Zero-shot=true2025.12 | 62.5 | |
| I-JEPAData=INet-22k, Arch.=g/162024.06 | 59.7 | |
| BEITData=INet-1k, Arch.=L/162024.06 | 59 | |
| CLIPPretrain=L-512M, Zero-shot=true2025.12 | 57.4 | |
| SuperCLIPPretrain=B-512M, Zero-shot=true2025.12 | 55.2 | |
| CLIPPretrain=B-512M, Zero-shot=true2025.12 | 53 | |
| AIMData=DFN-2B+, Arch.=H/142024.06 | 43.7 |