Image Classification on Caltech-256 (test)
87Top-1 AccOTTER
Evaluation results
| Method | Links | ||
|---|---|---|---|
| OTTERBackbone=ViT-B/16, Setting=Zero-shot2024.04 | 87 | — | |
| REGSLBackbone=ResNet-101, Pre-trained=ImageNet (ILSVRC-2012), Evaluation Protocol=Fine-tuning2021.11 | 84.9 | — | |
| l2-SPBackbone=ResNet-101, Pre-trained=ImageNet (ILSVRC-2012), Evaluation Protocol=Fine-tuning2021.11 | 84.13 | — | |
| CESamples per Class=full dataset2026.06 | 83.38 | — | |
| PolyLossSamples per Class=full dataset2026.06 | 83.34 | — | |
| l2-PGMBackbone=ResNet-101, Pre-trained=ImageNet (ILSVRC-2012), Evaluation Protocol=Fine-tuning2021.11 | 83.25 | — | |
| FR-ResNet18-Entropy#Samples=140002022.10 | 83.07 | — | |
| FR-ResNet18-Entropy#Samples=130002022.10 | 82.63 | — | |
| ResNet18-Entropy#Samples=140002022.10 | 82.61 | — | |
| FR-ResNet18-Entropy#Samples=120002022.10 | 82.44 | — | |
| ResNet18-Entropy#Samples=130002022.10 | 82.22 | — | |
| ResNet18-core-set#Samples=140002022.10 | 81.97 | — | |
| ResNet18-Entropy#Samples=120002022.10 | 81.82 | — | |
| ResNet18-core-set#Samples=130002022.10 | 81.62 | — | |
| FR-ResNet18-Entropy#Samples=110002022.10 | 81.6 | — | |
| TulaBatch-size=10242026.03 | 81.56 | — | |
| ResNet18-Entropy#Samples=110002022.10 | 81.37 | — | |
| FR-ResNet18-Entropy#Samples=100002022.10 | 81.34 | — | |
| PoLoBatch-size=10242026.03 | 81.14 | — | |
| LSBackbone=ResNet-101, Pre-trained=ImageNet (ILSVRC-2012), Evaluation Protocol=Fine-tuning2021.11 | 81.1 | — | |
| ResNet18-core-set#Samples=120002022.10 | 80.95 | — | |
| Distributional LossSamples per Class=full dataset, Density Estimate=KDE2026.06 | 80.91 | — | |
| ResNet18-core-set#Samples=110002022.10 | 80.65 | — | |
| ResNet18-Entropy#Samples=100002022.10 | 80.5 | — | |
| FR-ResNet18-Entropy#Samples=90002022.10 | 80.34 | — | |
| ResNet18-core-set#Samples=100002022.10 | 79.93 | — | |
| ResNet18-Entropy#Samples=90002022.10 | 79.81 | — | |
| Zero-shotBackbone=ViT-B/16, Setting=Zero-shot2024.04 | 79.8 | — | |
| TulaBatch-size=20482026.03 | 79.68 | — | |
| l2-NormBackbone=ResNet-101, Pre-trained=ImageNet (ILSVRC-2012), Evaluation Protocol=Fine-tuning2021.11 | 79.67 | — | |
| FR-ResNet18-Entropy#Samples=80002022.10 | 79.53 | — | |
| Fine-tuningBackbone=ResNet-101, Pre-trained=ImageNet (ILSVRC-2012), Evaluation Protocol=Fine-tuning2021.11 | 79.41 | — | |
| ResNet18-core-set#Samples=90002022.10 | 79.2 | — | |
| ResNet18-Entropy#Samples=80002022.10 | 78.89 | — | |
| LRSBatch-size=10242026.03 | 78.66 | — | |
| FR-ResNet18-Entropy#Samples=70002022.10 | 78.29 | — | |
| LLAL#Samples=140002022.10 | 78.14 | — | |
| ResNet18-core-set#Samples=80002022.10 | 78.12 | — | |
| LLAL#Samples=130002022.10 | 77.66 | — | |
| LLAL#Samples=120002022.10 | 77.59 | — | |
| ResNet18-Entropy#Samples=70002022.10 | 77.53 | — | |
| ResNet18-core-set#Samples=70002022.10 | 77.44 | — | |
| Proposed FrameworkBackbone=ResNet-502018.05 | 77.4 | — | |
| LLAL#Samples=110002022.10 | 77.12 | — | |
| FR-ResNet18-Entropy#Samples=60002022.10 | 77.11 | — | |
| LLAL#Samples=100002022.10 | 76.82 | — | |
| LBTBatch-size=10242026.03 | 76.8 | — | |
| LLAL#Samples=90002022.10 | 76.56 | — | |
| ResNet18-core-set#Samples=60002022.10 | 76.52 | — | |
| ResNet18-Entropy#Samples=60002022.10 | 76.03 | — | |
| LLAL#Samples=80002022.10 | 75.92 | — | |
| FR-ResNet18-Entropy#Samples=50002022.10 | 75.9 | — | |
| TulaBatch-size=40962026.03 | 75.7 | — | |
| PoLoBatch-size=20482026.03 | 75.54 | — | |
| T-vMFSamples per Class=full dataset2026.06 | 75.42 | — | |
| LLAL#Samples=70002022.10 | 75.21 | — | |
| CESamples per Class=202026.06 | 75.01 | — | |
| ResNet18-Entropy#Samples=50002022.10 | 74.79 | — | |
| ResNet18-core-set#Samples=50002022.10 | 74.79 | — | |
| LLAL#Samples=60002022.10 | 74.47 | — | |
| LARSBatch-size=40962026.03 | 74.29 | — | |
| Zeiler-Fergus2018.05 | 74.2 | — | |
| PolyLossSamples per Class=202026.06 | 74.08 | — | |
| LLAL#Samples=50002022.10 | 74.06 | — | |
| PoLoBatch-size=40962026.03 | 73.91 | — | |
| Distributional LossSamples per Class=20, Density Estimate=KDE2026.06 | 73.27 | — | |
| LARSBatch-size=20482026.03 | 72.82 | — | |
| SoftmaxClassifier=Softmax, Backbone=seven layers fixed model, Pre-training=ImageNet2018.05 | 72.6 | — | |
| SVMClassifier=SVM, Backbone=seven layers fixed model, Pre-training=ImageNet2018.05 | 71.7 | — | |
| LBTBatch-size=20482026.03 | 71.15 | — | |
| T-vMFSamples per Class=202026.06 | 69.37 | — | |
| Proposed FrameworkBackbone=VGG-162018.05 | 68.1 | — | |
| LARSBatch-size=10242026.03 | 67.64 | — | |
| LBTBatch-size=40962026.03 | 66.48 | — | |
| WaveMix-256/7Resolution=256 x 2562022.03 | 54.62 | — | |
| ResNet-18Resolution=256 x 2562022.03 | 52.97 | — | |
| ResNet-34Resolution=256 x 2562022.03 | 50.92 | — | |
| ResNet-50Resolution=256 x 2562022.03 | 49.97 | — | |
| T-vMFSamples per Class=22026.06 | 40.89 | — | |
| Distributional LossSamples per Class=2, Density Estimate=KDE2026.06 | 39.11 | — | |
| PolyLossSamples per Class=22026.06 | 38.32 | — | |
| CESamples per Class=22026.06 | 37.98 | — | |
| Image Codes2018.05 | 35.7 | — | |
| LRSBatch-size=20482026.03 | 10 | — | |
| Prior MatchingBackbone=ViT-B/16, Setting=Zero-shot2024.04 | 9.5 | — | |
| LRSBatch-size=40962026.03 | 5 | — | |
| AdaMixPublic Shots=5, ε=12022.03 | — | 23.74 | |
| AdaMixPublic Shots=5, ε=32022.03 | — | 20.85 | |
| Fully-PrivatePublic Shots=5, ε=12022.03 | — | 60.72 | |
| Fully-PrivatePublic Shots=5, ε=32022.03 | — | 27.15 | |
| Non-PrivatePublic Shots=5, Setting=paragon2022.03 | — | 14.64 | |
| Only-PublicPublic Shots=5, Setting=baseline2022.03 | — | 24.88 | |
| PPGDPublic Shots=5, ε=12022.03 | — | 24.61 | |
| PPGDPublic Shots=5, ε=32022.03 | — | 23.71 |