Image Classification on MIT-67 (MIT-Indoor) (test)
79.3Top-1 AccREGSL
Evaluation results
| Method | Links | ||
|---|---|---|---|
| REGSLBackbone=ResNet-101, Pre-trained=ImageNet (ILSVRC-2012), Evaluation Protocol=Fine-tuning2021.11 | 79.3 | — | |
| l2-SPBackbone=ResNet-101, Pre-trained=ImageNet (ILSVRC-2012), Evaluation Protocol=Fine-tuning2021.11 | 78.11 | — | |
| l2-PGMBackbone=ResNet-101, Pre-trained=ImageNet (ILSVRC-2012), Evaluation Protocol=Fine-tuning2021.11 | 77.31 | — | |
| LSBackbone=ResNet-101, Pre-trained=ImageNet (ILSVRC-2012), Evaluation Protocol=Fine-tuning2021.11 | 76.84 | — | |
| FixMatch +AKC+ARC#label=13402021.03 | 76.64 | — | |
| l2-NormBackbone=ResNet-101, Pre-trained=ImageNet (ILSVRC-2012), Evaluation Protocol=Fine-tuning2021.11 | 76.57 | — | |
| MixMatch+AKC+ARC#label=13402021.03 | 75.54 | — | |
| Fine-tuningBackbone=ResNet-101, Pre-trained=ImageNet (ILSVRC-2012), Evaluation Protocol=Fine-tuning2021.11 | 74.78 | — | |
| LwFBackbone=VGG-16, Resolution=224x2242022.03 | 74.78 | — | |
| FinetuningBackbone=VGG-16, Resolution=224x2242022.03 | 74.7 | — | |
| EWCBackbone=VGG-16, Resolution=224x224, lambda=0.5k2022.03 | 74.7 | — | |
| FixMatch#label=13402021.03 | 74.27 | — | |
| AKC+ARC#label=13402021.03 | 73.31 | — | |
| MixMatch#label=13402021.03 | 73.14 | — | |
| ARC#label=13402021.03 | 72.72 | — | |
| EWCBackbone=VGG-16, Resolution=224x224, lambda=8k2022.03 | 72.69 | — | |
| AKC#label=13402021.03 | 71.93 | — | |
| Pseudo label#label=13402021.03 | 71.68 | — | |
| Mean teacher#label=13402021.03 | 71.34 | — | |
| FixMatch +AKC+ARC#label=6702021.03 | 70.61 | — | |
| MixMatch+AKC+ARC#label=6702021.03 | 70.3 | — | |
| Supervised labeled#label=13402021.03 | 68.94 | — | |
| MixMatch#label=6702021.03 | 68.58 | — | |
| FixMatch#label=6702021.03 | 68.31 | — | |
| AKC+ARC#label=6702021.03 | 67.44 | — | |
| ARC#label=6702021.03 | 66.94 | — | |
| AKC#label=6702021.03 | 66.64 | — | |
| Mean teacher#label=6702021.03 | 64.37 | — | |
| Pseudo label#label=6702021.03 | 63.77 | — | |
| Supervised labeled#label=6702021.03 | 63.35 | — | |
| LwFBackbone=ResNet-18, Evaluation Protocol=Observed Accuracy2022.03 | 57.6 | — | |
| Finetuning (SupCon)Backbone=ResNet-18, Evaluation Protocol=Observed Accuracy, Loss Function=Supervised Contrastive2022.03 | 57.1 | — | |
| Finetuning (Cross-Entropy)Backbone=ResNet-18, Evaluation Protocol=Observed Accuracy, Loss Function=Cross-Entropy2022.03 | 56.9 | — | |
| EWCBackbone=ResNet-18, Evaluation Protocol=Observed Accuracy, lambda=0.5k2022.03 | 52.5 | — | |
| MixMatch+AKC+ARC#label=1342021.03 | 48.54 | — | |
| FixMatch +AKC+ARC#label=1342021.03 | 48.34 | — | |
| AKC+ARC#label=1342021.03 | 47.11 | — | |
| AKC#label=1342021.03 | 46.79 | — | |
| ARC#label=1342021.03 | 46.67 | — | |
| MixMatch#label=1342021.03 | 44.65 | — | |
| Supervised labeled#label=1342021.03 | 44.28 | — | |
| FixMatch#label=1342021.03 | 44.13 | — | |
| Mean teacher#label=1342021.03 | 43.05 | — | |
| EWCBackbone=ResNet-18, Evaluation Protocol=Observed Accuracy, lambda=8k2022.03 | 42.1 | — | |
| Pseudo label#label=1342021.03 | 39.28 | — | |
| AdaLNBackbone=ViT-L/162023.03 | — | 15.1 | |
| AdaLNBackbone=SWIN-L, % Trainable parameters=0.1%, No backbone backpropagation=false2023.03 | — | 8.9 | |
| AdaMixPublic Shots=5, ε=12022.03 | — | 40.46 | |
| AdaMixPublic Shots=5, ε=32022.03 | — | 33.03 | |
| Attention Transfer (ResNet-18)Type=AT, Backbone=ResNet-18, Teacher=ResNet-34, Pre-training=ImageNet, Evaluation Protocol=Finetuning2016.12 | — | 27.1 | |
| BitFitBackbone=ViT-L/162023.03 | — | 15.1 | |
| BitFitBackbone=SWIN-L, % Trainable parameters=0.1%, No backbone backpropagation=false2023.03 | — | 8.5 | |
| Full FTBackbone=ViT-L/162023.03 | — | 10.4 | |
| Full FTBackbone=SWIN-L, % Trainable parameters=100%, No backbone backpropagation=false2023.03 | — | 10.5 | |
| Fully-PrivatePublic Shots=5, ε=12022.03 | — | 69.55 | |
| Fully-PrivatePublic Shots=5, ε=32022.03 | — | 42.99 | |
| In. LPBackbone=ViT-L/162023.03 | — | 9.7 | |
| In. LPBackbone=SWIN-L, % Trainable parameters=0.1%, No backbone backpropagation=true2023.03 | — | 9.9 | |
| In. MLP-3Backbone=ViT-L/162023.03 | — | 10.1 | |
| In. MLP-3Backbone=SWIN-L, % Trainable parameters=2.8%, No backbone backpropagation=true2023.03 | — | 10.2 | |
| InCABackbone=ViT-L/162023.03 | — | 9 | |
| InCABackbone=SWIN-L, % Trainable parameters=3.7%, No backbone backpropagation=true2023.03 | — | 10.1 | |
| InCA (last)Backbone=ViT-L/162023.03 | — | 9 | |
| InCA (last)Backbone=SWIN-L, % Trainable parameters=3.7%, No backbone backpropagation=true2023.03 | — | 10.1 | |
| KD (ResNet-18)Type=KD, Backbone=ResNet-18, Teacher=ResNet-34, Pre-training=ImageNet, Evaluation Protocol=Finetuning2016.12 | — | 28.1 | |
| LoRABackbone=ViT-L/162023.03 | — | 14.8 | |
| LoRABackbone=SWIN-L, % Trainable parameters=0.8%, No backbone backpropagation=false2023.03 | — | 9.6 | |
| LPBackbone=ViT-L/162023.03 | — | 10.5 | |
| LPBackbone=SWIN-L, % Trainable parameters=0.1%, No backbone backpropagation=false2023.03 | — | 10.3 | |
| MLP-3Backbone=ViT-L/162023.03 | — | 11.2 | |
| MLP-3Backbone=SWIN-L, % Trainable parameters=2.8%, No backbone backpropagation=false2023.03 | — | 10.2 | |
| Non-PrivatePublic Shots=5, Setting=paragon2022.03 | — | 25.17 | |
| Only-PublicPublic Shots=5, Setting=baseline2022.03 | — | 43.58 | |
| PPGDPublic Shots=5, ε=12022.03 | — | 42.02 | |
| PPGDPublic Shots=5, ε=32022.03 | — | 39.6 | |
| ResNet-18 StudentType=student, Backbone=ResNet-18, Pre-training=ImageNet, Evaluation Protocol=Finetuning2016.12 | — | 28.2 | |
| ResNet-34 TeacherType=teacher, Backbone=ResNet-34, Pre-training=ImageNet, Evaluation Protocol=Finetuning2016.12 | — | 26 | |
| VPTBackbone=ViT-L/162023.03 | — | 14.8 |