List your company
ResearchFormerly WizWand

Image Classification on Aircraft (Top-1 Accuracy)

57.4Top-1 AccOurs
-1.25613.97229.244.428Feb 15, 2022Nov 4, 2022Jul 24, 2023Apr 11, 2024Dec 29, 2024Sep 17, 2025Jun 6, 2026
Updated 4mo ago

Evaluation results

MethodLinks
OursShots=16, Backbone=ViT-B/16, Seeds=32025.1257.4
IsoCLIPIntra-modal=✓, Classifier=NCM, Backbone=ViT-B/16-open2026.0354.5
Image-ImageIntra-modal=✓, Classifier=NCM, Backbone=ViT-B/16-open2026.0354
OSTBProtocol=Final Ensemble2026.0652.24
OursShots=8, Backbone=ViT-B/16, Seeds=32025.1249.7
DST (FixMatch)Pre-training=Unsupervised, Backbone=ResNet-50, Labels per category=42022.0247.7
DST (FlexMatch)Pre-training=Unsupervised, Backbone=ResNet-50, Labels per category=42022.0247.3
DST (FlexMatch)Pre-training=Supervised, Backbone=ResNet-50, Labels per category=42022.0244.5
DST (FixMatch)Pre-training=Supervised, Backbone=ResNet-50, Labels per category=42022.0243.2
CARPEp=400, k=20, Evaluation Protocol=k-NN, Multi-crop=false2023.1042.1
LlipData Size=2.5B, Vision Encoder=ViT-B/16, Evaluation Protocol=Zero-shot2024.1241.5
OursShots=4, Backbone=ViT-B/16, Seeds=32025.1240.2
MoCo-v3Ep=1000, k=20, Evaluation Protocol=k-NN, Multi-crop=false2023.1036.9
DINOEp=800, k=20, Evaluation Protocol=k-NN, Multi-crop=true2023.1035.3
OursShots=2, Backbone=ViT-B/16, Seeds=32025.1234.9
CARPEp=400, k=20, Evaluation Protocol=k-NN, Multi-crop=true2023.1034.8
TripletEp=980, k=20, Evaluation Protocol=k-NN, Multi-crop=false2023.1033.4
Barlow TwinsEp=1000, k=20, Evaluation Protocol=k-NN, Multi-crop=false2023.1032.7
OursShots=1, Backbone=ViT-B/16, Seeds=32025.1232.2
DeepCluster-v2Ep=800, k=20, Evaluation Protocol=k-NN, Multi-crop=true2023.1032
MetaCLIPData Size=2.5B, Vision Encoder=ViT-B/16, Evaluation Protocol=Zero-shot2024.1230.3
Image-TextIntra-modal=✗, Classifier=Zero-Shot, Backbone=ViT-B/16-open2026.0329.8
SwAVEp=800, k=20, Evaluation Protocol=k-NN, Multi-crop=true2023.1029
BaselinePre-training=Unsupervised, Backbone=ResNet-50, Labels per category=42022.0228.7
OpenCLIPData Size=2B, Vision Encoder=ViT-B/16, Evaluation Protocol=Zero-shot2024.1225.8
BaselinePre-training=Supervised, Backbone=ResNet-50, Labels per category=42022.0225.4
CLIP-ViT-B/16Layers=-2026.0324.33
TCA (Baseline)Layers=3, 6, 9, KeepRate=0.72026.0323.82
Col-LnLayers=3, 6, 9, KeepRate=0.72026.0322.98
Col-LnLayers=0, 3, 6, KeepRate=0.72026.0322.17
TCA (Baseline)Layers=0, 3, 6, KeepRate=0.72026.0322.05
SeLa-v2Ep=400, k=20, Evaluation Protocol=k-NN, Multi-crop=true2023.1020.7
CLIP-B/32Backbone=CLIP-B/32, Protocol=Zero-shot2026.0619.17
InfoMinEp=800, k=20, Evaluation Protocol=k-NN, Multi-crop=false2023.1018.2
oBoWEp=200, k=20, Evaluation Protocol=k-NN, Multi-crop=true2023.1018.1
DreamLIPData Size=30M, Vision Encoder=ViT-B/16, Evaluation Protocol=Zero-shot2024.127.9
FLAIRData Size=30M, Vision Encoder=ViT-B/16, Evaluation Protocol=Zero-shot2024.127.2
CLIPData Size=30M, Vision Encoder=ViT-B/16, Evaluation Protocol=Zero-shot2024.125.7
SigLIPData Size=30M, Vision Encoder=ViT-B/16, Evaluation Protocol=Zero-shot2024.125.7
HyCoCLIPw/ boxes=true2025.103.57
MERUw/ boxes=false2025.103.53
MERUw/ boxes=true2025.103.53
PHyCLIPw/ boxes=true2025.103.24
CLIP-PGSImage Encoder=ViT-B/16, p=0.3, Protocol=Zero-shot2025.033.1
CLIPw/ boxes=false2025.103.01
CLIPw/ boxes=true2025.102.89
CLIP-PGSImage Encoder=ViT-S/16, p=0.3, Protocol=Zero-shot2025.032.7
E-CLIPImage Encoder=ViT-B/16, Protocol=Zero-shot2025.032.2
CLIP-PGSImage Encoder=ViT-S/16, p=0.5, Protocol=Zero-shot2025.032
CLIP-PGSImage Encoder=ViT-B/16, p=0.5, Protocol=Zero-shot2025.031.9
FLAIRData Size=3M, Vision Encoder=ViT-B/16, Evaluation Protocol=Zero-shot2024.121.7
CLIPImage Encoder=ViT-B/16, Protocol=Zero-shot2025.031.7
LaCLIPData Size=3M, Vision Encoder=ViT-B/16, Evaluation Protocol=Zero-shot2024.121.6
DreamLIPData Size=3M, Vision Encoder=ViT-B/16, Evaluation Protocol=Zero-shot2024.121.6
MLLM-AData Size=3M, Vision Encoder=ViT-B/16, Evaluation Protocol=Zero-shot2024.121.5
SigLIPData Size=3M, Vision Encoder=ViT-B/16, Evaluation Protocol=Zero-shot2024.121.4
FLIPImage Encoder=ViT-B/16, Protocol=Zero-shot2025.031.4
A-CLIPImage Encoder=ViT-B/16, Protocol=Zero-shot2025.031.3
CLIPData Size=3M, Vision Encoder=ViT-B/16, Evaluation Protocol=Zero-shot2024.121