List your company
ResearchFormerly WizWand

Image Classification on Oxford-IIIT Pets

97.9AccuracyBASIC-L
93.32494.51295.796.888May 28, 2019Jul 22, 2020Sep 16, 2021Nov 11, 2022Jan 6, 2024Mar 2, 2025Apr 27, 2026
Updated 4mo ago

Evaluation results

MethodLinks
BASIC-LBackbone=BASIC-L2021.1197.9
SoViT-400m/14N-shot=10, Probe=linear regression2023.0597.6
ViT-g/14N-shot=10, Probe=linear regression2023.0597.4
LaCLIPArchitecture=ViT-B/16, Pre-training Data=LAION-400M, Protocol=5-way 5-shot2023.0597.1
ViT-L/16N-shot=10, Probe=linear regression2023.0597
StatABackbone=ViT-L/14, Scenario=High, Batch Size=1282025.0196.5
StatABackbone=ViT-L/14, Batch Size=128, Online Test-Time Adaptation=true, Scenario=High2025.0196.5
DINOv2Architecture=ViT-g/14, Pre-training Data=LVD, Resolution=518x518, Evaluation Protocol=Attentive Probing, Backbone State=Frozen2024.0196.4
StatABackbone=ViT-L/14, Scenario=Medium, Batch Size=1282025.0196.3
StatABackbone=ViT-L/14, Scenario=Separate, Batch Size=1282025.0196.3
StatABackbone=ViT-L/14, Batch Size=128, Online Test-Time Adaptation=true, Scenario=Medium2025.0196.3
StatABackbone=ViT-L/14, Batch Size=128, Online Test-Time Adaptation=true, Scenario=Separate2025.0196.3
CLIPArchitecture=ViT-B/16, Pre-training Data=LAION-400M, Protocol=5-way 5-shot2023.0596.2
GPipeNumber of Parameters=556M, Pre-training=ImageNet, Transfer Learning=true2019.0595.9
Florence-CoSwin-HBackbone=CoSwin-H, Resolution=384px, Evaluation Protocol=Zero-shot2021.1195.9
AmoebaNet-B (6,512)2019.0695.9
EVA-02-CLIP-E/14+zero-shot evaluation protocol=true2023.0395.8
LaCLIPArchitecture=ViT-B/32, Pre-training Data=LAION-400M, Protocol=5-way 5-shot2023.0595.7
Mini-DeiT-B↑384#Params=44M2022.0495.5
EfficientNet-B6Number of Parameters=41M, Pre-training=ImageNet, Transfer Learning=true2019.0595.4
EfficientNet-B7# Params=64.0M, # FLOPs=37.2B, Resolution=600, Pre-training=ImageNet, Fine-tuning=true2021.0795.4
AIM-7B†Architecture=ViT-7B/14, Pre-training Data=DFN-2B+, Feature Extraction Layer=20th, Evaluation Protocol=Attentive Probing, Backbone State=Frozen2024.0195.4
BASIC-MBackbone=BASIC-M2021.1195.3
MLCDPre-training Data=LAION-400M, Backbone=ViT-L/14, Evaluation Protocol=Linear Probe2024.0795.3
CMT-S# Params=25.1M, # FLOPs=4.04B, Pre-training=ImageNet, Fine-tuning=true2021.0795.2
MAEArchitecture=ViT-2B/14, Pre-training Data=IG-3B, Evaluation Protocol=Attentive Probing, Backbone State=Frozen2024.0195.1
CLIPPre-training Data=WIT-400M, Backbone=ViT-L/14, Evaluation Protocol=Linear Probe2024.0795.1
ViT-BBackbone=ViT-B, ForAug=true, Pre-training Dataset=ImageNet, Evaluation Protocol=Finetuning2025.0395.1
ViT-LBackbone=ViT-L, ForAug=true, Pre-training Dataset=ImageNet, Evaluation Protocol=Finetuning2025.0395.1
MAEArchitecture=ViT-H/14, Pre-training Data=IN-1k, Evaluation Protocol=Attentive Probing, Backbone State=Frozen2024.0195
StatABackbone=ViT-L/14, Scenario=Low, Batch Size=1282025.0195
StatABackbone=ViT-L/14, Batch Size=128, Online Test-Time Adaptation=true, Scenario=Low2025.0195
TGR-MoEModel Size=Base, Top-K Routing=22026.0494.97
CeiT-SFLOPs=12.9G, Resolution=384x384, Pre-training=ImageNet2021.0394.9
CeiT-S# Params=24.2M, # FLOPs=12.9B, Resolution=384, Pre-training=ImageNet, Fine-tuning=true2021.0794.9
CLIP+Pre-training Data=WIT-400M, Backbone=ViT-L/14, Evaluation Protocol=Linear Probe2024.0794.9
Swin-SBackbone=Swin-S, ForAug=true, Pre-training Dataset=ImageNet, Evaluation Protocol=Finetuning2025.0394.9
EfficientNet-B4Number of Parameters=17M, Pre-training=ImageNet, Transfer Learning=true2019.0594.8
FixResBackbone=SENet-1542019.0694.8
Swin-TiBackbone=Swin-Ti, ForAug=true, Pre-training Dataset=ImageNet, Evaluation Protocol=Finetuning2025.0394.8
TNT-S# Params=23.8M, # FLOPs=17.3B, Resolution=384, Pre-training=ImageNet, Fine-tuning=true2021.0794.7
OURSShots=8, Backbone=ViT-B/16, Templates=72025.1294.7
Full Fine-tuningBackbone=ViT-L/16, Pre-trained=ImageNet12K2026.0494.67
TGR-MoEModel Size=Base, Top-K Routing=12026.0494.63
OpenCLIP ViT-H/14Zero-shot=true, Backbone=ViT-H/14, Confidence Interval=95% CI, Runs=52025.1094.61
CeiT-SFLOPs=4.5G, Resolution=224x224, Pre-training=ImageNet2021.0394.6
SENet-154Evaluation Protocol=Baseline2019.0694.6
OpenCLIP VIT-H/14zero-shot=true2023.0394.6
ResNet-152Number of Parameters=58M, Pre-training=ImageNet, Transfer Learning=true2019.0594.5
CeiT-TFLOPs=3.6G, Resolution=384x384, Pre-training=ImageNet2021.0394.5
TWISTEvaluation Protocol=Fine-tune, Backbone=ResNet-50, Pre-trained=ImageNet2021.1094.5
ResNet-152# Params=58.1M, # FLOPs=11.3B, Pre-training=ImageNet, Fine-tuning=true2021.0794.5
CLIPArchitecture=ViT-B/32, Pre-training Data=LAION-400M, Protocol=5-way 5-shot2023.0594.5
CLIPArchitecture=ViT-B/16, Pre-training Data=RedCaps, Protocol=5-way 5-shot2023.0594.5
ViT-SBackbone=ViT-S, ForAug=true, Pre-training Dataset=ImageNet, Evaluation Protocol=Finetuning2025.0394.5
Swin-SBackbone=Swin-S, ForAug=false, Pre-training Dataset=ImageNet, Evaluation Protocol=Finetuning2025.0394.4
ResNet-101Backbone=ResNet-101, ForAug=false, Pre-training Dataset=ImageNet, Evaluation Protocol=Finetuning2025.0394.3
VMoE w/ z-lossModel Size=Base, Top-K Routing=12026.0494.21
ViTAE-SParams (M)=23.62022.0294.2
UNICOMPre-training Data=LAION-400M, Backbone=ViT-L/14, Evaluation Protocol=Linear Probe2024.0794.2
OURSShots=8, Backbone=ViT-B/16, Templates=12025.1294.2
ViT-LBackbone=ViT-L, ForAug=false, Pre-training Dataset=ImageNet, Evaluation Protocol=Finetuning2025.0394.2
Swin-TiBackbone=Swin-Ti, ForAug=false, Pre-training Dataset=ImageNet, Evaluation Protocol=Finetuning2025.0394.2
TGR-MoEModel Size=Small, Top-K Routing=22026.0494.14
LaCLIPArchitecture=ViT-B/16, Pre-training Data=RedCaps, Protocol=5-way 5-shot2023.0594.1
AIM-3BArchitecture=ViT-3B/14, Pre-training Data=DFN-2B+, Evaluation Protocol=Attentive Probing, Backbone State=Frozen2024.0194.1
ViT-BBackbone=ViT-B, ForAug=false, Pre-training Dataset=ImageNet, Evaluation Protocol=Finetuning2025.0394.1
BaLoRABackbone=ViT-L/16, Pre-trained=ImageNet12K2026.0494.07
SAM + GAMBackbone=Swin-t, Evaluation Protocol=finetuning2023.0394.03
DoRABackbone=ViT-L/16, Pre-trained=ImageNet12K2026.0494.01
Skip TuningShot=16, Time (s)=962, Mem. (M)=5282024.1294
ResNet-101Backbone=ResNet-101, ForAug=true, Pre-training Dataset=ImageNet, Evaluation Protocol=Finetuning2025.0394
CRATE-αModel variant=L/142024.0593.98
LoRABackbone=ViT-L/16, Pre-trained=ImageNet12K2026.0493.96
AdamW + GAMBackbone=Swin-t, Evaluation Protocol=finetuning2023.0393.87
ViT-B/16FLOPs=18.7G, Pre-training=ImageNet2021.0393.8
CeiT-TFLOPs=1.2G, Resolution=224x224, Pre-training=ImageNet2021.0393.8
ViT-B/16#Params=86M2022.0493.8
ViT-B/16# Params=85.8M, # FLOPs=17.6B, Resolution=384, Pre-training=ImageNet, Fine-tuning=true2021.0793.8
ViT-B/16Params (M)=86.52022.0293.8
AIM-7BArchitecture=ViT-7B/14, Pre-training Data=DFN-2B+, Evaluation Protocol=Attentive Probing, Backbone State=Frozen2024.0193.8
ViT-SBackbone=ViT-S, ForAug=false, Pre-training Dataset=ImageNet, Evaluation Protocol=Finetuning2025.0393.8
TGR-MoEModel Size=Small, Top-K Routing=12026.0493.79
VMoEModel Size=Base, Top-K Routing=12026.0493.79
Inception-v4# Params=41.1M, # FLOPs=16.1B, Pre-training=ImageNet, Fine-tuning=true2021.0793.7
OURSShots=4, Backbone=ViT-B/16, Templates=12025.1293.7
ResNet-50Backbone=ResNet-50, ForAug=false, Pre-training Dataset=ImageNet, Evaluation Protocol=Finetuning2025.0393.7
ResNet-50Backbone=ResNet-50, ForAug=true, Pre-training Dataset=ImageNet, Evaluation Protocol=Finetuning2025.0393.7
Expert Choice MoEModel Size=Small, Top-K Routing=12026.0493.69
ProSRCShot=16, Time (s)=14403, Mem. (M)=33732024.1293.67
ViT (dense)Model Size=Base, Top-K Routing=12026.0493.65
ViT-L/16FLOPs=65.8G, Pre-training=ImageNet2021.0393.6
ViT-L/16Params (M)=304.32022.0293.6
SAMBackbone=Swin-t, Evaluation Protocol=finetuning2023.0393.59
CLIP-ViT-L/14Backbone=ViT-L/14, Resolution=336px, Evaluation Protocol=Zero-shot2021.1193.5
CLIPBackbone=ViT-L/14-3362021.1193.5
CLIPBackbone=ViT-L/14, Setting=Zero-shot, Batch Size=1282025.0193.5
CLIPBackbone=ViT-L/14, Batch Size=128, Online Test-Time Adaptation=false2025.0193.5
ProSRCShot=8, Time (s)=7636, Mem. (M)=33732024.1293.5
OURSShots=4, Backbone=ViT-B/16, Templates=72025.1293.5