List your company
ResearchFormerly WizWand

Image Classification on Oxford 102 Flowers

99.6Top-1 AccuracyHUVR
56.12867.41478.789.986Jun 14, 2019Aug 11, 2020Oct 9, 2021Dec 7, 2022Feb 4, 2024Apr 3, 2025Jun 1, 2026
Updated 5mo ago

Evaluation results

MethodLinks
HUVREncoder=ViT-B, Dimension=32, Compression=HUVR, Evaluation Protocol=Linear probing2026.0199.6
HUVREncoder=ViT-L, Dimension=32, Compression=HUVR, Evaluation Protocol=Linear probing2026.0199.5
HUVREncoder=ViT-B, Dimension=16, Compression=HUVR, Evaluation Protocol=Linear probing2026.0199.4
SigLIP 2Encoder=ViT-L, Dimension=32, Compression=PCA, Evaluation Protocol=Linear probing2026.0199.3
Core-tuningBackbone=ResNet-50, Pre-training=MoCo-v22021.0299.18
DINOv3Encoder=ViT-L, Dimension=32, Compression=PCA, Evaluation Protocol=Linear probing2026.0199.1
DeiT-B⚗ ↑384#Params=88M2022.0498.9
HUVREncoder=ViT-B, Dimension=8, Compression=HUVR, Evaluation Protocol=Linear probing2026.0198.9
EfficientNet-B72019.0698.8
EfficientNet-B7#Params=66M2022.0498.8
DeiT-B⚗#Params=87M2022.0498.8
SL-CE-tuningBackbone=ResNet-50, Pre-training=Supervised2021.0298.78
RIFLEBackbone=ResNet-50, Pre-training=MoCo-v22021.0298.7
L2SPBackbone=ResNet-50, Pre-training=MoCo-v22021.0298.66
DELTABackbone=ResNet-50, Pre-training=MoCo-v22021.0298.65
SCLBackbone=ResNet-50, Pre-training=MoCo-v22021.0298.65
SigLIP 2Encoder=ViT-B, Dimension=32, Compression=PCA, Evaluation Protocol=Linear probing2026.0198.6
C-RADIOv3Encoder=ViT-L, Dimension=32, Compression=PCA, Evaluation Protocol=Linear probing2026.0198.6
M&MBackbone=ResNet-50, Pre-training=MoCo-v22021.0298.57
BSSBackbone=ResNet-50, Pre-training=MoCo-v22021.0298.57
Bi-tuningBackbone=ResNet-50, Pre-training=MoCo-v22021.0298.57
EfficientNet-B5#Params=30M2022.0498.5
DeiT-B↑384#Params=87M2022.0498.5
CE-tuningBackbone=ResNet-50, Pre-training=MoCo-v22021.0298.49
DeiT-B#Params=86M2022.0498.4
Mini-DeiT-B↑384#Params=44M2022.0498.3
Grafit ResNet-50#Params=25M2022.0498.2
AWTShots=162024.0797.69
SUPEvaluation Protocol=Fine-tune, Backbone=ResNet-50, Pre-trained=ImageNet2021.1097.6
DINOv3Encoder=ViT-B, Dimension=32, Compression=PCA, Evaluation Protocol=Linear probing2026.0197.3
TWISTEvaluation Protocol=Fine-tune, Backbone=ResNet-50, Pre-trained=ImageNet2021.1097.1
CoOpShots=162024.0797.07
BYOLEvaluation Protocol=Fine-tune, Backbone=ResNet-50, Pre-trained=ImageNet2021.1097
OURSShots=8, Backbone=ViT-B/16, Templates=12025.1296.9
SimCLREvaluation Protocol=Fine-tune, Backbone=ResNet-50, Pre-trained=ImageNet2021.1096.6
CLIP-LoRAShots=8, Backbone=ViT-B/16, Templates=12025.1296.3
RényiCLEvaluation Protocol=linear evaluation, Multi-crop Augmentation=true2022.0896.2
OURSShots=8, Backbone=ViT-B/16, Templates=72025.1296.2
BYOLEvaluation Protocol=Linear evaluation, Backbone=ResNet-50, Pre-trained=ImageNet2021.1096.1
CLIP-LoRAShots=8, Backbone=ViT-B/16, Templates=72025.1295.8
FixResBackbone=InceptionResNet-V22019.0695.7
InceptionResNet-V2Evaluation Protocol=Baseline2019.0695
OURSShots=4, Backbone=ViT-B/16, Templates=72025.1294.9
OURSShots=4, Backbone=ViT-B/16, Templates=12025.1294.8
SUPEvaluation Protocol=Linear evaluation, Backbone=ResNet-50, Pre-trained=ImageNet2021.1094.7
DINOBackbone=ViT-B/16, Pre-training Epochs=400, Training Heuristics=w/ heuristics, Linear Probe Training Epochs=102026.0694.6
RényiCLEvaluation Protocol=linear evaluation, Multi-crop Augmentation=false2022.0894.2
iBOTBackbone=ViT-L/16, Pre-training Epochs=250, Training Heuristics=w/ heuristics, Linear Probe Training Epochs=102026.0694
LoRA-CLIPShots=4, Backbone=ViT-B/16, Templates=72025.1293.9
C-RADIOv3Encoder=ViT-B, Dimension=32, Compression=PCA, Evaluation Protocol=Linear probing2026.0193.7
LoRA-CLIPShots=4, Backbone=ViT-B/16, Templates=12025.1293.7
iBOTBackbone=ViT-B/16, Pre-training Epochs=400, Training Heuristics=w/ heuristics, Linear Probe Training Epochs=102026.0693.7
TWISTEvaluation Protocol=Linear evaluation, Backbone=ResNet-50, Pre-trained=ImageNet2021.1093.4
ViT-BAdaptation Strategy=Fine-tuning, Params=85.8M, Pre-training Dataset=ImageNet-1K2026.0592.9
bViT-BAdaptation Strategy=LoRA, FFN, Params=62K, Pre-training Dataset=ImageNet-1K2026.0592.6
bViT-BAdaptation Strategy=Fine-tuning, Params=7.8M, Pre-training Dataset=ImageNet-1K2026.0592.3
VISRegBackbone=ViT-L/14, Pre-training Epochs=100, Training Heuristics=w/o heuristics, Linear Probe Training Epochs=1002026.0692.3
RandomEvaluation Protocol=Fine-tune, Backbone=ResNet-50, Pre-trained=ImageNet2021.1092
ViT-BAdaptation Strategy=LoRA, QV, Params=296K, Pre-training Dataset=ImageNet-1K2026.0591.9
ViT-BAdaptation Strategy=LoRA, FFN, Params=738K, Pre-training Dataset=ImageNet-1K2026.0591.8
bViT-BAdaptation Strategy=LoRA, QV, Params=25K, Pre-training Dataset=ImageNet-1K2026.0591.7
MoCoV3Backbone=ViT-B/16, Pre-training Epochs=300, Training Heuristics=w/ heuristics, Linear Probe Training Epochs=102026.0691.5
SimCLREvaluation Protocol=Linear evaluation, Backbone=ResNet-50, Pre-trained=ImageNet2021.1091.2
InfoMinEvaluation Protocol=linear evaluation, Multi-crop Augmentation=false2022.0891.2
CLSAEvaluation Protocol=linear evaluation, Multi-crop Augmentation=true2022.0891.2
LeJEPA*Backbone=ViT-L/14, Pre-training Epochs=100, Training Heuristics=w/o heuristics, Linear Probe Training Epochs=1002026.0691.2
bViT-BAdaptation Strategy=Time embedding tuning, Params=9K, Pre-training Dataset=ImageNet-1K2026.0591.1
CLSAEvaluation Protocol=linear evaluation, Multi-crop Augmentation=false2022.0890.8
VISRegBackbone=ViT-B/16, Pre-training Epochs=400, Training Heuristics=w/o heuristics, Linear Probe Training Epochs=102026.0690.4
VISRegBackbone=ViT-L/14, Pre-training Epochs=400, Training Heuristics=w/o heuristics, Linear Probe Training Epochs=102026.0690.2
ViT-B/16#Params=86M2022.0489.5
CoCoOpShots=162024.0787.84
VP (white-box)Shots=16, Model Access Type=White-box2024.0786.9
I-JEPABackbone=ViT-H/14, Pre-training Epochs=300, Training Heuristics=w/ heuristics, Linear Probe Training Epochs=102026.0686.8
DINOv3Encoder=ViT-B, Dimension=16, Compression=PCA, Evaluation Protocol=Linear probing2026.0186.7
bViT-BAdaptation Strategy=Linear probing, Params=—, Pre-training Dataset=ImageNet-1K2026.0586.7
AWTShots=12024.0785.61
ViT-B/32#Params=86M2022.0485.4
MAEBackbone=ViT-L/16, Pre-training Epochs=1600, Training Heuristics=w/o heuristics, Linear Probe Training Epochs=102026.0685.4
LoRA-CLIPShots=1, Backbone=ViT-B/16, Templates=72025.1283.7
OURSShots=1, Backbone=ViT-B/16, Templates=12025.1283.3
LoRA-CLIPShots=1, Backbone=ViT-B/16, Templates=12025.1283.2
ViT-BAdaptation Strategy=Linear probing, Params=—, Pre-training Dataset=ImageNet-1K2026.0583
OURSShots=1, Backbone=ViT-B/16, Templates=72025.1282.8
data2vecBackbone=ViT-L/14, Pre-training Epochs=1600, Training Heuristics=w/ heuristics, Linear Probe Training Epochs=102026.0681.4
CoOpShots=12024.0777.53
CoCoOpShots=12024.0772.08
BlackVIP-SEShots=16, Model Access Type=Black-box2024.0771.5
ZSEvaluation Protocol=Zero-shot2024.0771.3
BARShots=16, Model Access Type=Black-box2024.0771.2
BlackVIPShots=16, Model Access Type=Black-box2024.0770.6
Handcrafted2024.0767.44
VP w/ SPSA-GCShots=16, Model Access Type=Black-box2024.0767
DINOv3Encoder=ViT-B, Dimension=8, Compression=PCA, Evaluation Protocol=Linear probing2026.0157.8