Image Classification on Oxford Flowers (test)
99.65AccuracySAM-final
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| SAM-finalBackbone=EfficientNet-L2, Fine-tuning strategy=With SAM optimization2021.02 | 99.65 | — | — | — | |
| ALIGNBackbone=EfficientNet-L2, Resolution=289/3602021.02 | 99.65 | — | — | — | |
| BiT-LBackbone=ResNet152 x 42021.02 | 99.63 | — | — | — | |
| SAM-baselineBackbone=EfficientNet-L2, Fine-tuning strategy=Without SAM optimization2021.02 | 99.6 | — | — | — | |
| CAMALModel=MaxViT2026.05 | 98.8 | — | — | — | |
| AttentiveNAS-A6MFLOPS=7092020.11 | 98.6 | — | — | — | |
| PriorModel=MaxViT2026.05 | 98.6 | — | — | — | |
| PriorModel=Swin2026.05 | 98.1 | — | — | — | |
| EfficientNet-B1MFLOPS=10502020.11 | 97.6 | — | — | — | |
| AttentiveNAS-A1MFLOPS=2792020.11 | 97.4 | — | — | — | |
| CAMALModel=Swin2026.05 | 97.3 | — | — | — | |
| EfficientNet-B0MFLOPS=3902020.11 | 96.9 | — | — | — | |
| PriorModel=EfficientNet2026.05 | 96.1 | — | — | — | |
| MoCo-v2Backbone=ResNet-50, Pre-training=ImageNet-1K, Evaluation Setting=Fine-tuning, Public Checkpoint=true2021.12 | 95.62 | — | — | — | |
| CAMALModel=EfficientNet2026.05 | 95.6 | — | — | — | |
| SupervisedBackbone=ResNet-50, Evaluation Setting=Fine-tuning2021.12 | 95.5 | — | — | — | |
| MMCLBackbone=ResNet-50, Pre-training=ImageNet-1K, Evaluation Setting=Fine-tuning2021.12 | 95.24 | — | — | — | |
| MOCHIBackbone=ResNet-50, Pre-training=ImageNet-1K, Evaluation Setting=Fine-tuning, Public Checkpoint=true2021.12 | 94.8 | — | — | — | |
| PriorModel=ViT2026.05 | 94.2 | — | — | — | |
| PCL-v2Backbone=ResNet-50, Pre-training=ImageNet-1K, Evaluation Setting=Fine-tuning2021.12 | 92.95 | — | — | — | |
| CAMALModel=ConvNeXt2026.05 | 92.9 | — | — | — | |
| CAMALModel=ViT2026.05 | 92.5 | — | — | — | |
| CAMALModel=ResNet2026.05 | 92.3 | — | — | — | |
| PriorModel=ConvNeXt2026.05 | 91.7 | — | — | — | |
| Gated Prompt Tuning#OF PTS=48, Backbone=ViT-S, Pre-training=MoCo v32023.06 | 91.14 | — | — | — | |
| PCL-v1Backbone=ResNet-50, Pre-training=ImageNet-1K, Evaluation Setting=Fine-tuning2021.12 | 90.83 | — | — | — | |
| MLRSTM2025.12 | 90.28 | — | — | — | |
| LSQMM2025.12 | 90.28 | — | — | — | |
| VPT-Deep#OF PTS=48, Backbone=ViT-S, Pre-training=MoCo v32023.06 | 89.53 | — | — | — | |
| InsDisBackbone=ResNet-50, Pre-training=ImageNet-1K, Evaluation Setting=Fine-tuning2021.12 | 89.51 | — | — | — | |
| PriorModel=ResNet2026.05 | 89.5 | — | — | — | |
| MoCoBackbone=ResNet-50, Pre-training=ImageNet-1K, Evaluation Setting=Fine-tuning2021.12 | 89.45 | — | — | — | |
| BitFitBackbone=ResNet-50, Pre-trained=true2026.02 | 88.5 | — | — | — | |
| SMM2025.12 | 87.95 | — | — | — | |
| Model projectionBackbone=ResNet-50, Pre-trained=true2026.02 | 87.58 | — | — | — | |
| Model projection + F.T.Backbone=ResNet-50, Pre-trained=true2026.02 | 87.51 | — | — | — | |
| Batch NormalizationBackbone=ResNet-50, Pre-trained=true2026.02 | 87.48 | — | — | — | |
| 2-step model projectionBackbone=ResNet-50, Pre-trained=true2026.02 | 87.45 | — | — | — | |
| LRBackbone=ResNet-50, Pre-trained=true2026.02 | 87.28 | — | — | — | |
| TT-MMK2025.12 | 86.59 | — | — | — | |
| LIBSVM2025.12 | 86.51 | — | — | — | |
| VPT-Shallow#OF PTS=48, Backbone=ViT-S, Pre-training=MoCo v32023.06 | 84.65 | — | — | — | |
| LiT-L16L + RECO# params (M)=652, zero-shot=true2023.06 | 84.1 | — | — | — | |
| LR + FTBackbone=ResNet-50, Pre-trained=true2026.02 | 82.94 | — | — | — | |
| HopSBackbone=R18, Ambiguity level (q)=0.67, Prompting Strategy=uni-prompt2026.04 | 82.5 | — | — | — | |
| HopSBackbone=R50, Ambiguity level (q)=0.67, Prompting Strategy=uni-prompt2026.04 | 82 | — | — | — | |
| LoRA-r8Backbone=ResNet-50, Pre-trained=true, rank=82026.02 | 80.96 | — | — | — | |
| PaRABackbone=ResNet-50, Pre-trained=true2026.02 | 80.65 | — | — | — | |
| CLIP-L/14 + RECO# params (M)=435, zero-shot=true2023.06 | 79.5 | — | — | — | |
| K-Lite# params (M)=151, zero-shot=true2023.06 | 78.6 | — | — | — | |
| LoRA-r32Backbone=ResNet-50, Pre-trained=true, rank=322026.02 | 78.57 | — | — | — | |
| CroSelBackbone=R50, Ambiguity level (q)=0.67, Prompting Strategy=uni-prompt2026.04 | 78 | — | — | — | |
| LiT-L16L# params (M)=638, zero-shot=true2023.06 | 77.4 | — | — | — | |
| CLIP-L/14# params (M)=428, zero-shot=true2023.06 | 75.6 | — | — | — | |
| HopSBackbone=R18, Ambiguity level (q)=0.75, Prompting Strategy=uni-prompt2026.04 | 75.1 | — | — | — | |
| CroSelBackbone=R18, Ambiguity level (q)=0.67, Prompting Strategy=uni-prompt2026.04 | 75 | — | — | — | |
| HopSBackbone=R50, Ambiguity level (q)=0.75, Prompting Strategy=uni-prompt2026.04 | 74 | — | — | — | |
| HopSBackbone=CLIP, Ambiguity level (q)=0.67, Prompting Strategy=uni-prompt2026.04 | 73.9 | — | — | — | |
| HopSBackbone=R18, Ambiguity level (q)=0.80, Prompting Strategy=uni-prompt2026.04 | 69.9 | — | — | — | |
| CLIP-B/32 + RECO# params (M)=154, zero-shot=true2023.06 | 67.9 | — | — | — | |
| PapiBackbone=R18, Ambiguity level (q)=0.67, Prompting Strategy=uni-prompt2026.04 | 67.9 | — | — | — | |
| PapiBackbone=R50, Ambiguity level (q)=0.67, Prompting Strategy=uni-prompt2026.04 | 67 | — | — | — | |
| HopSBackbone=R50, Ambiguity level (q)=0.80, Prompting Strategy=uni-prompt2026.04 | 65 | — | — | — | |
| Align# params (M)=247, zero-shot=true2023.06 | 64.9 | — | — | — | |
| HopSBackbone=CLIP, Ambiguity level (q)=0.75, Prompting Strategy=uni-prompt2026.04 | 64.8 | — | — | — | |
| CLIP-B/32# params (M)=151, zero-shot=true2023.06 | 62.1 | — | — | — | |
| CroSelBackbone=R50, Ambiguity level (q)=0.75, Prompting Strategy=uni-prompt2026.04 | 62 | — | — | — | |
| HopSBackbone=CLIP, Ambiguity level (q)=0.80, Prompting Strategy=uni-prompt2026.04 | 59.2 | — | — | — | |
| PapiBackbone=R50, Ambiguity level (q)=0.75, Prompting Strategy=uni-prompt2026.04 | 58.7 | — | — | — | |
| CroSelBackbone=R18, Ambiguity level (q)=0.75, Prompting Strategy=uni-prompt2026.04 | 56.4 | — | — | — | |
| CLIP-R-50 + RECO# params (M)=114, zero-shot=true2023.06 | 56.2 | — | — | — | |
| PapiBackbone=R18, Ambiguity level (q)=0.75, Prompting Strategy=uni-prompt2026.04 | 56.1 | — | — | — | |
| FTBackbone=ResNet-50, Pre-trained=true2026.02 | 52.81 | — | — | — | |
| RA-CLIP# params (M)=151, zero-shot=true2023.06 | 52.3 | — | — | — | |
| CroSelBackbone=R50, Ambiguity level (q)=0.80, Prompting Strategy=uni-prompt2026.04 | 50.2 | — | — | — | |
| PapiBackbone=R50, Ambiguity level (q)=0.80, Prompting Strategy=uni-prompt2026.04 | 49.8 | — | — | — | |
| CLIP-R-50# params (M)=102, zero-shot=true2023.06 | 47.2 | — | — | — | |
| PapiBackbone=R18, Ambiguity level (q)=0.80, Prompting Strategy=uni-prompt2026.04 | 47 | — | — | — | |
| CroSelBackbone=R18, Ambiguity level (q)=0.80, Prompting Strategy=uni-prompt2026.04 | 45.1 | — | — | — | |
| CroSelBackbone=CLIP, Ambiguity level (q)=0.67, Prompting Strategy=uni-prompt2026.04 | 40.4 | — | — | — | |
| PapiBackbone=CLIP, Ambiguity level (q)=0.67, Prompting Strategy=uni-prompt2026.04 | 35.7 | — | — | — | |
| CroSelBackbone=CLIP, Ambiguity level (q)=0.75, Prompting Strategy=uni-prompt2026.04 | 27.4 | — | — | — | |
| PapiBackbone=CLIP, Ambiguity level (q)=0.75, Prompting Strategy=uni-prompt2026.04 | 13.7 | — | — | — | |
| CroSelBackbone=CLIP, Ambiguity level (q)=0.80, Prompting Strategy=uni-prompt2026.04 | 11.5 | — | — | — | |
| PapiBackbone=CLIP, Ambiguity level (q)=0.80, Prompting Strategy=uni-prompt2026.04 | 5.8 | — | — | — | |
| AdaMixPublic Shots=2, ε=12022.03 | — | 39.48 | — | — | |
| AdaMixPublic Shots=2, ε=32022.03 | — | 36.11 | — | — | |
| CCUPModel=ViT-32, Source Dataset=StanfordDogs, Target Classes=Pekinese, toy poodle, Scotch terrier2025.12 | — | — | 70.6 | 69.3 | |
| CCUPModel=ViT-32, Source Dataset=StanfordCars, Target Classes=2009 Spyker C8 Coupe, 2010 Dodge Ram Pickup 3500 Crew Cab, 2011 Ford Ranger SuperCab2025.12 | — | — | 70.6 | 70.2 | |
| CCUPModel=ViT-32, Source Dataset=Caltech101, Target Classes=euphonium, minaret, platypus2025.12 | — | — | 70.6 | 70.1 | |
| CCUPModel=ViT-16, Source Dataset=StanfordDogs, Target Classes=Pekinese, toy poodle, Scotch terrier2025.12 | — | — | 70.2 | 68.6 | |
| CCUPModel=ViT-16, Source Dataset=StanfordCars, Target Classes=2009 Spyker C8 Coupe, 2010 Dodge Ram Pickup 3500 Crew Cab, 2011 Ford Ranger SuperCab2025.12 | — | — | 70.2 | 68.1 | |
| CCUPModel=ViT-16, Source Dataset=Caltech101, Target Classes=euphonium, minaret, platypus2025.12 | — | — | 70.2 | 68.5 | |
| Fully-PrivatePublic Shots=2, ε=12022.03 | — | 95.12 | — | — | |
| Fully-PrivatePublic Shots=2, ε=32022.03 | — | 81.42 | — | — | |
| LipModel=RN50, Source Dataset=StanfordDogs, Target Classes=Pekinese, toy poodle, Scotch terrier2025.12 | — | — | 66.1 | 63.3 | |
| LipModel=RN50, Source Dataset=StanfordCars, Target Classes=2009 Spyker C8 Coupe, 2010 Dodge Ram Pickup 3500 Crew Cab, 2011 Ford Ranger SuperCab2025.12 | — | — | 66.1 | 60.7 | |
| LipModel=RN50, Source Dataset=Caltech101, Target Classes=euphonium, minaret, platypus2025.12 | — | — | 66.1 | 63.3 | |
| LipModel=ViT-B/16, Source Dataset=StanfordDogs, Target Classes=Pekinese, toy poodle, Scotch terrier2025.12 | — | — | 70.8 | 71.3 | |
| LipModel=ViT-B/16, Source Dataset=StanfordCars, Target Classes=2009 Spyker C8 Coupe, 2010 Dodge Ram Pickup 3500 Crew Cab, 2011 Ford Ranger SuperCab2025.12 | — | — | 70.8 | 69.9 |