Image Classification on iNaturalist 2019
88.5Top-1 AccHiera-H448
Evaluation Results
| Method | Links | |
|---|---|---|
| Hiera-H448backbone=Hiera-H4482023.06 | 88.5 | |
| MAEBackbone=ViT-H, Input Resolution=448, Pre-train Data=IN1K, Evaluation Protocol=Fine-tuned2021.11 | 88.3 | |
| ViT-H448backbone=ViT-H4482023.06 | 88.3 | |
| MAEBackbone=ViT-H, Pre-train Data=IN1K, Evaluation Protocol=Fine-tuned2021.11 | 85.7 | |
| ViT-Hbackbone=ViT-H2023.06 | 85.7 | |
| Hiera-Hbackbone=Hiera-H2023.06 | 85.7 | |
| MixMAEBackbone=Swin-L, FLOPs (G)=35.8, Params. (M)=2352022.05 | 84.4 | |
| Hiera-Lbackbone=Hiera-L2023.06 | 84.3 | |
| GrafitBackbone=RegNetY-8.0GF, Pre-training=ImageNet-1k, Resolution=384x384, Number of Parameters=39M, Classifier=MLP2020.11 | 84.1 | |
| GrafitBackbone=RegNetY-8.0GF, Parameters=39M, Resolution=384x3842020.11 | 84.1 | |
| Previous best2021.11 | 84.1 | |
| RDNet-LParam (M)=186, FLOPs (G)=34.7, Pre-training=ImageNet-1k2024.03 | 83.7 | |
| RDNet-BParam (M)=87, FLOPs (G)=15.4, Pre-training=ImageNet-1k2024.03 | 83.5 | |
| MAEBackbone=ViT-L, Pre-train Data=IN1K, Evaluation Protocol=Fine-tuned2021.11 | 83.4 | |
| MAEBackbone=ViT-L, FLOPs (G)=61.3, Params. (M)=3042022.05 | 83.4 | |
| ViT-Lbackbone=ViT-L2023.06 | 83.4 | |
| MixMAEBackbone=Swin-B/W14, FLOPs (G)=16.3, Params. (M)=882022.05 | 83.3 | |
| Hiera-B+backbone=Hiera-B+2023.06 | 83.1 | |
| DeiT-B⚗ ↑384im/sec=85.9, resolution=384, distillation=true2020.12 | 83 | |
| Hiera-Bbackbone=Hiera-B2023.06 | 83 | |
| CeiT-SFLOPs=12.9G, Resolution=384x384, Pre-training=ImageNet2021.03 | 82.7 | |
| RDNet-SParam (M)=50, FLOPs (G)=8.7, Pre-training=ImageNet-1k2024.03 | 82.4 | |
| CaiT-M-36 ↑ 224Resolution=224, Crop-ratio=0.875, Training=Longer schedules2021.03 | 81.8 | |
| CaiT-M-36 224Resolution=224, Crop-ratio=0.8752021.03 | 81.7 | |
| DeiT-B ↑384im/sec=85.9, resolution=3842020.12 | 81.4 | |
| CaiT-S-36 ↑ 224Resolution=224, Crop-ratio=0.875, Training=Longer schedules2021.03 | 81.4 | |
| Deit-BFLOPs=52.8G, Resolution=384x384, Pre-training=ImageNet2021.03 | 81.4 | |
| RDNet-TParam (M)=24, FLOPs (G)=5.0, Pre-training=ImageNet-1k2024.03 | 81.2 | |
| MugsBackbone=ViT-B/162022.03 | 80.8 | |
| CaiT-S-36 224Resolution=224, Crop-ratio=0.8752021.03 | 80.6 | |
| MAEBackbone=ViT-B, Pre-train Data=IN1K, Evaluation Protocol=Fine-tuned2021.11 | 80.5 | |
| MAEBackbone=ViT-B/162022.03 | 80.5 | |
| MAEBackbone=ViT-B, FLOPs (G)=17.5, Params. (M)=862022.05 | 80.5 | |
| ViT-Bbackbone=ViT-B2023.06 | 80.5 | |
| Grafit RegNetY-8GFim/sec=591.62020.12 | 80 | |
| Grafit RegNetY-8GFFLOPs=8.0G, Pre-training=ImageNet2021.03 | 80 | |
| MugsBackbone=ViT-S/162022.03 | 79.8 | |
| iBOTBackbone=ViT-B/162022.03 | 79.6 | |
| DeiT-III-LParam (M)=304, FLOPs (G)=61.6, Pre-training=ImageNet-1k2024.03 | 79.3 | |
| BEITBackbone=ViT-B/162022.03 | 79.2 | |
| CeiT-SFLOPs=4.5G, Resolution=224x224, Pre-training=ImageNet2021.03 | 78.9 | |
| DINOBackbone=ViT-B/16, Pre-training=Self-supervised, Evaluation Protocol=finetuning2021.04 | 78.6 | |
| DINOBackbone=ViT-B/162022.03 | 78.6 | |
| iBOTBackbone=ViT-S/162022.03 | 78.5 | |
| DeiT-B⚗im/sec=290.9, resolution=224, distillation=true2020.12 | 78.4 | |
| DINOBackbone=ViT-S/16, Pre-training=Self-supervised, Evaluation Protocol=finetuning2021.04 | 78.2 | |
| DINOBackbone=ViT-S/162022.03 | 78.2 | |
| DINOBackbone=ViT-B, FLOPs (G)=17.5, Params. (M)=862022.05 | 78.2 | |
| DINOEvaluation Protocol=Fine-Tuning, Backbone=ViT-Base/16, Pre-training=ImageNet-1K2022.04 | 78.2 | |
| MSNEvaluation Protocol=Fine-Tuning, Backbone=ViT-Base/16, Pre-training=ImageNet-1K2022.04 | 78.1 | |
| DeiT-III-BParam (M)=87, FLOPs (G)=17.5, Pre-training=ImageNet-1k2024.03 | 78 | |
| GrowTAS-S#Params (M)=22.42025.12 | 77.94 | |
| CeiT-TFLOPs=3.6G, Resolution=384x384, Pre-training=ImageNet2021.03 | 77.9 | |
| DeiT-Bim/sec=292.3, resolution=2242020.12 | 77.7 | |
| DeiT-B 224Resolution=224, Crop-ratio=0.8752021.03 | 77.7 | |
| Deit-BFLOPs=17.3G, Resolution=224x224, Pre-training=ImageNet2021.03 | 77.7 | |
| SupervisedBackbone=ViT-B/16, Pre-training=Supervised, Evaluation Protocol=finetuning2021.04 | 77.7 | |
| Deit-B/16FLOPs=17.5B, Resolution=224, Pre-training Dataset=ImageNet-1k, Evaluation Protocol=Fine-tune2021.05 | 77.7 | |
| Sup.Backbone=ViT-B/162022.03 | 77.7 | |
| DeiT-BParams (M)=86.62022.02 | 77.7 | |
| DeiT-BParam (M)=87, FLOPs (G)=17.5, Pre-training=ImageNet-1k2024.03 | 77.7 | |
| DeiT-B#Params (M)=86.02025.12 | 77.7 | |
| DeiT-BaseParams (M)=86.5, FLOPs (G)=17.56, Structured Compression=false, Fine-tuning=true2025.05 | 77.7 | |
| AttnScale=Base, Params (M)=86.52026.05 | 77.7 | |
| FiPS+RigL FTBackbone=DeiT-B, P. Budget=50%, Comp. Ratio=(∼25%)2024.11 | 77.69 | |
| FiPS+RigL FTBackbone=DeiT-B, P. Budget=40%, Comp. Ratio=(∼33%)2024.11 | 77.58 | |
| GFMBackbone=DeiT-B, P. Budget=50%, Comp. Ratio=(∼25%)2024.11 | 77.56 | |
| DINO + SEREMode=Fine-tuning2022.06 | 77.5 | |
| AutoFormer-S#Params (M)=22.92025.12 | 77.4 | |
| OriginalBackbone=DeiT-B, P. Budget=100, Comp. Ratio=–2024.11 | 77.39 | |
| FiPS+RigL FTBackbone=DeiT-B, P. Budget=25%, Comp. Ratio=(∼45%)2024.11 | 77.26 | |
| GFMBackbone=DeiT-B, P. Budget=40%, Comp. Ratio=(∼33%)2024.11 | 77.13 | |
| NViT-STask Mode=Fine-tuning, Pre-training=ImageNet2021.10 | 77 | |
| DINOMode=Fine-tuning2022.06 | 76.9 | |
| BiMambaScale=Base, Params (M)=89.92026.05 | 76.9 | |
| ViT-Slimps#Params (M)=15.6, FLOPs (B)=3.32022.01 | 76.67 | |
| SupervisedBackbone=ViT-S/16, Pre-training=Supervised, Evaluation Protocol=finetuning2021.04 | 76.6 | |
| Sup.Backbone=ViT-S/162022.03 | 76.6 | |
| BEITBackbone=ViT-S/162022.03 | 76.5 | |
| PreNAS-S#Params (M)=23.02025.12 | 76.4 | |
| ViTAE-SParams (M)=23.62022.02 | 76 | |
| ViTAE-S#Params (M)=24.02025.12 | 76 | |
| GrafitBackbone=ResNet-50, Pre-training=ImageNet, Resolution=224x224, Evaluation Protocol=Single center crop, Classifier=MLP2020.11 | 75.9 | |
| Grafit ResNet-50im/sec=1226.12020.12 | 75.9 | |
| Grafit ResNet-50FLOPs=4.1G, Pre-training=ImageNet2021.03 | 75.9 | |
| Grafit ResNet-50Params (M)=25.62022.02 | 75.9 | |
| Grafit ResNet-50Param (M)=26, FLOPs (G)=4.1, Pre-training=ImageNet-1k2024.03 | 75.9 | |
| DEIT-S#Params (M)=22.0, FLOPs (B)=4.62022.01 | 75.35 | |
| FAR-BaseParams (M)=83.2, FLOPs (G)=17.31, Structured Compression=false, Fine-tuning=true2025.05 | 75.1 | |
| BiLSTMScale=Base, Params (M)=83.22026.05 | 75.1 | |
| NViT-TTask Mode=Fine-tuning, Pre-training=ImageNet2021.10 | 74.65 | |
| Grafit FCBackbone=ResNet-50, Pre-training=ImageNet, Resolution=224x224, Evaluation Protocol=Single center crop, Classifier=Linear FC2020.11 | 74.6 | |
| Grafit/ResNet50FLOPs=4.1B, Resolution=224, Pre-training Dataset=ImageNet-1k, Evaluation Protocol=Fine-tune2021.05 | 74.6 | |
| SNCA+Backbone=ResNet-50, Pre-training=ImageNet, Resolution=224x224, Evaluation Protocol=Single center crop, Classifier=MLP2020.11 | 74.5 | |
| FAR-Base (Pruned)Params (M)=73.0, FLOPs (G)=14.80, Structured Compression=true, Fine-tuning=true2025.05 | 74.4 | |
| BiMambaScale=Small, Params (M)=26.92026.05 | 74.3 | |
| DeiT-STask Mode=Fine-tuning, Pre-training=ImageNet2021.10 | 74.22 | |
| DeiT-SmallParams (M)=22.1, FLOPs (G)=4.60, Structured Compression=false, Fine-tuning=true2025.05 | 74.2 | |
| AttnScale=Small, Params (M)=22.92026.05 | 74.2 | |
| FAR-SmallParams (M)=23.9, FLOPs (G)=4.82, Structured Compression=false, Fine-tuning=true2025.05 | 73.9 |