List your company
ResearchFormerly WizWand

Image Classification on Oxford-IIIT Pet (Top-1 accuracy)

95.9Top-1 AccuracyPiSSA
47.43660.01872.685.182Mar 27, 2022Dec 1, 2022Aug 8, 2023Apr 14, 2024Dec 20, 2024Aug 27, 2025May 4, 2026
Updated 6mo ago

Evaluation results

MethodLinks
PiSSABackbone=ViT-Base, Budget=100%2026.0595.9
PiSSABackbone=ViT-Large, Budget=100%2026.0595.7
SigLIP2Data size=10B, Zero-shot=true2026.0395.4
JACTUSBackbone=ViT-Large, Budget=80%2026.0595.2
LoRABackbone=ViT-Large, Budget=100%2026.0594.8
DoRABackbone=ViT-Large, Budget=100%2026.0594.8
PEcoreData size=5B, Zero-shot=true2026.0394.6
AdapterTuneBackbone=ViT-B/16, Trainable Parameters (%)=0.9%2026.0394.3
SigLIPData size=10B, Zero-shot=true2026.0394.2
SVD-LLM+LoRABackbone=ViT-Large, Budget=80%2026.0594.2
DoRABackbone=ViT-Base, Budget=100%2026.0593.8
AdapterTuneBackbone=ViT-S/16, Trainable Parameters (%)=0.9%2026.0393.5
LlipData size=2.5B, Zero-shot=true2026.0393.5
JACTUSBackbone=ViT-Base, Budget=80%2026.0593.4
Head-only tuningBackbone=ViT-B/16, Trainable Parameters (%)=0.1%2026.0393.3
SVFTBackbone=ViT-Large, Budget=100%2026.0593.3
JACTUSBackbone=ViT-Large, Budget=60%2026.0593.2
LoRABackbone=ViT-Base, Budget=100%2026.0593.1
SVD-LLM+LoRABackbone=ViT-Base, Budget=80%2026.0592.7
CapPaEvaluation Protocol=10-shot linear evaluation, Model Scale=L/142023.0692.6
SVFTBackbone=ViT-Base, Budget=100%2026.0592.5
SVD-LLM+LoRABackbone=ViT-Large, Budget=60%2026.0592.2
CaCoBackbone=ResNet-50, Pre-training Dataset=ImageNet-1K, Pre-training Epochs=8002022.0391.9
NNCLRBackbone=ResNet-50, Pre-training Dataset=ImageNet-1K, Pre-training Epochs=8002022.0391.8
IN-RealFakeBackbone=ViT-S, Pre-Trained Data=IN-RealFake, Evaluation Protocol=Linear Probing2024.1291.7
MetaCLIPData size=2.5B, Zero-shot=true2026.0391.7
JACTUSBackbone=ViT-Base, Budget=60%2026.0591.6
AdapterTuneBackbone=DeiT-T, Trainable Parameters (%)=0.9%2026.0391.4
Head-only tuningBackbone=DeiT-T, Trainable Parameters (%)=0.1%2026.0390.8
IN-RealBackbone=ResNet50, Pre-Trained Data=IN-Real, Evaluation Protocol=Linear Probing2024.1290.7
Head-only tuningBackbone=ViT-S/16, Trainable Parameters (%)=0.1%2026.0390.5
BYOLBackbone=ResNet-50, Pre-training Dataset=ImageNet-1K, Pre-training Epochs=8002022.0390.4
CLIPEvaluation Protocol=10-shot linear evaluation, Model Scale=L/142023.0690.4
IN-RealBackbone=ViT-S, Pre-Trained Data=IN-Real, Evaluation Protocol=Linear Probing2024.1290.2
VP (white-box)Shots=16, Model Access Type=White-box2024.0790.2
SVD-LLM+LoRABackbone=ViT-Base, Budget=60%2026.0590.2
IN-RealFakeBackbone=ResNet50, Pre-Trained Data=IN-RealFake, Evaluation Protocol=Linear Probing2024.1290
BlackVIP-SEShots=16, Model Access Type=Black-box2024.0789.8
BlackVIPShots=16, Model Access Type=Black-box2024.0789.7
Full fine-tuningBackbone=ViT-S/16, Trainable Parameters (%)=100%2026.0389.5
OpenCLIPData size=2B, Zero-shot=true2026.0389.5
JACTUSBackbone=ViT-Large, Budget=40%2026.0589.4
ZSEvaluation Protocol=Zero-shot2024.0789.1
Full fine-tuningBackbone=DeiT-T, Trainable Parameters (%)=100%2026.0389
lbGenBackbone=ViT-S, Pre-Trained Data=IN-lbGen, Evaluation Protocol=Linear Probing2024.1288.6
BARShots=16, Model Access Type=Black-box2024.0788.6
JACTUSBackbone=ViT-Base, Budget=40%2026.0588.2
CLIP*Evaluation Protocol=10-shot linear evaluation, Model Scale=L/142023.0687.7
MoCo v2+Backbone=ResNet-50, Pre-training Dataset=ImageNet-1K, Pre-training Epochs=8002022.0387.6
lbGenBackbone=ResNet50, Pre-Trained Data=IN-lbGen, Evaluation Protocol=Linear Probing2024.1287.2
VP w/ SPSA-GCShots=16, Model Access Type=Black-box2024.0787.1
AdCo +Backbone=ResNet-50, Pre-training Dataset=ImageNet-1K, Pre-training Epochs=8002022.0386.8
Full fine-tuningBackbone=ViT-B/16, Trainable Parameters (%)=100%2026.0386.6
CapPaEvaluation Protocol=10-shot linear evaluation2023.0686.5
SVD-LLM+LoRABackbone=ViT-Large, Budget=40%2026.0585.8
IN-SD1.5Backbone=ViT-S, Pre-Trained Data=IN-SD1.5, Evaluation Protocol=Linear Probing2024.1285.6
IN-SD1.5Backbone=ResNet50, Pre-Trained Data=IN-SD1.5, Evaluation Protocol=Linear Probing2024.1285.5
SimCLRBackbone=ResNet-50, Pre-training Dataset=ImageNet-1K, Pre-training Epochs=8002022.0384.6
CapEvaluation Protocol=10-shot linear evaluation2023.0683.7
SVD-LLM+LoRABackbone=ViT-Base, Budget=40%2026.0583.4
IN-GenRobustBackbone=ResNet50, Pre-Trained Data=IN-GenRobust, Evaluation Protocol=Linear Probing2024.1283.3
IN-GenRobustBackbone=ViT-S, Pre-Trained Data=IN-GenRobust, Evaluation Protocol=Linear Probing2024.1282.1
CLIPEvaluation Protocol=10-shot linear evaluation2023.0682.1
CLIP*Evaluation Protocol=10-shot linear evaluation, Batch Size=16k2023.0680.6
CLIP*Evaluation Protocol=10-shot linear evaluation, Batch Size=8k2023.0677.7
Noun SubmanifoldBackbone=CLIP ViT-B-32, Evaluation Protocol=Zero-shot2023.0575.2
POS PGABackbone=CLIP ViT-B-32, Evaluation Protocol=Zero-shot2023.0575.1
CLIPBackbone=CLIP ViT-B-32, Evaluation Protocol=Zero-shot2023.0573.8
POS PCABackbone=CLIP ViT-B-32, Evaluation Protocol=Zero-shot2023.0573.8
DreamLIPData size=30M, Zero-shot=true2026.0364.1
COSMOSData size=30M, Zero-shot=true2026.0362.9
GoldiCLIPData size=30M, Zero-shot=true2026.0362.7
FLAIRData size=30M, Zero-shot=true2026.0355.6
SigLIPData size=30M, Zero-shot=true2026.0353.3
CLIPData size=30M, Zero-shot=true2026.0349.3