List your company
ResearchFormerly WizWand

Few-shot Image Classification on CIFAR-FS (test) (1-shot and 5-shot)

94.87AccuracySeedSelect
53.405264.170174.93585.6999Jul 21, 2020Jan 13, 2021Jul 8, 2021Jan 1, 2022Jun 26, 2022Dec 19, 2022Jun 14, 2023
Updated 8mo ago

Evaluation results

MethodLinks
SeedSelectFew-shot setting=5-way 5-shot2023.0694.87——
NAO-centroidFew-shot setting=5-way 5-shot2023.0694.85——
DiffAlignFew-shot setting=5-way 5-shot, pre-trained on external datasets=true2023.0691.96——
Stable DiffusionFew-shot setting=5-way 5-shot, pre-trained on external datasets=true2023.0690.87——
FeLMiFew-shot setting=5-way 5-shot2023.0689.47——
Label-HallucFew-shot setting=5-way 5-shot2023.0689.37——
Sparse and Shared Feature ActivationsArchitecture=ResNet12, Shots=52023.0486.9——
MCRNet-SVMbackbone=ResNet-12, shots=5-shot, base-learner=SVM2020.0786.8——
SEGAFew-shot setting=5-way 5-shot, multi-modal (class labels)=true2023.0686——
MetaBaselineArchitecture=WRN 28-10, Shots=52023.0485.79——
RFSshots=52021.0885.7——
MeLashots=52021.0885.6——
MCRNet-RRbackbone=ResNet-12, shots=5-shot, base-learner=Ridge Regression2020.0785.2——
Shot-freeshots=52021.0884.7——
Baseline-RRbackbone=ResNet-12, shots=5-shot, base-learner=Ridge Regression2020.0784.3——
MetaOptNetshots=52021.0884.3——
Baseline-SVMbackbone=ResNet-12, shots=5-shot, base-learner=SVM2020.0784.2——
MetaOptNetArchitecture=ResNet12, Shots=52023.0484.2——
ProtoNetshots=52021.0883.5——
Fine-tuningbackbone=ResNet-12, shots=5-shot2020.0782.13——
R2D2backbone=4CONV, shots=5-shot2020.0779.4——
R2D2shots=52021.0879.4——
SVAEFew-shot setting=5-way 5-shot, pre-trained on external datasets=true2023.0678.89——
MetaBaselineArchitecture=WRN 28-10, Shots=12023.0476.58——
Sparse and Shared Feature ActivationsArchitecture=ResNet12, Shots=12023.0475.1——
MCRNet-SVMbackbone=ResNet-12, shots=1-shot, base-learner=SVM2020.0774.7——
MCRNet-RRbackbone=ResNet-12, shots=1-shot, base-learner=Ridge Regression2020.0773.8——
Baseline-RRbackbone=ResNet-12, shots=1-shot, base-learner=Ridge Regression2020.0772.6——
MetaOptNetshots=12021.0872.6——
ProtoNetshots=12021.0872.2——
Prototypical Networksbackbone=4CONV, shots=5-shot2020.0772——
Baseline-SVMbackbone=ResNet-12, shots=1-shot, base-learner=SVM2020.0772——
MetaOptNetArchitecture=ResNet12, Shots=12023.0472——
RFSshots=12021.0871.6——
MAMLbackbone=4CONV, shots=5-shot2020.0771.5——
MAMLshots=52021.0871.5——
MeLashots=12021.0871.4——
Relation Networksbackbone=4CONV, shots=5-shot2020.0769.3——
Shot-freeshots=12021.0869.2——
R2D2backbone=4CONV, shots=1-shot2020.0765.3——
R2D2shots=12021.0865.3——
Fine-tuningbackbone=ResNet-12, shots=1-shot2020.0764.66——
MAMLbackbone=4CONV, shots=1-shot2020.0758.9——
MAMLshots=12021.0858.9——
Prototypical Networksbackbone=4CONV, shots=1-shot2020.0755.5——
Relation Networksbackbone=4CONV, shots=1-shot2020.0755——
DistillBackbone=ResNet-12, Params=12.5M, FLOPs=3.5 × 10^92023.03—73.986.9
EfficientFSLBackbone=ViT-S, Params=1.25M, Pre-training=ImageNet1K2026.01—88.8294.6
EfficientFSLBackbone=ViT-B, Params=2.48M, Pre-training=ImageNet1K2026.01—85.2592.64
EfficientFSLBackbone=ViT-B, Params=2.48M, Pre-training=ImageNet21K2026.01—93.2597.28
FewTUREBackbone=ViT-S, Params=21.7M2026.01—72.886.14
FewVSBackbone=ViT-S, Params=21.7M2026.01—85.6390.73
Full Fine-TuningBackbone=ViT-S, Params=21.7M2026.01—84.8692.3
Full Fine-TuningBackbone=ViT-B, Params=85.8M2026.01—85.8392.88
KTPPBackbone=Visformer-T, Params=11.1M2026.01—83.6390.19
LRSetting=Inductive, Backbone=ResNet-12, Input Size=84x842020.03—62.2580.82
LR + ICISetting=Transductive, Backbone=ResNet-12, Input Size=84x842020.03—73.9784.13
LR + ICISetting=Semi-supervised, Backbone=ResNet-12, Input Size=84x84, Unlabeled Samples=15/152020.03—73.6783.85
LR + ICISetting=Semi-supervised, Backbone=ResNet-12, Input Size=84x84, Unlabeled Samples=30/502020.03—76.5184.32
LR + ICISetting=Semi-supervised, Backbone=ResNet-12, Input Size=84x84, Unlabeled Samples=80/802020.03—78.0784.76
MABASBackbone=ResNet-12, Params=12.5M, FLOPs=3.5 × 10^92023.03—73.5185.49
MetaF.Backbone=ViT-S, Params=24.5M2026.01—88.3492.21
MetaOptNetSetting=Inductive, Backbone=ResNet-12, Input Size=84x842020.03—72.684.3
MetaOptNetBackbone=ResNet-12, Params=12.5M, FLOPs=3.5 × 10^92023.03—72.684.3
PN+rotBackbone=WRN-28-10, Params=36.5M, FLOPs=3.7 × 10^102023.03—69.5582.34
Pre-trainBackbone=Visformer-T, Params=10.0M, FLOPs=1.3 × 10^92023.03—71.9985.98
ProtoNetSetting=Inductive, Backbone=ResNet-18, Input Size=224x2242020.03—72.283.5
ProtoNetBackbone=ResNet-12, Params=12.5M, FLOPs=3.5 × 10^92023.03—72.283.5
RE-NetBackbone=ResNet-12, Params=12.5M, FLOPs=3.5 × 10^92023.03—74.5186.6
SCAM-NetBackbone=ViT-S, Params=21.7M2026.01—79.4590.9
SemFewBackbone=Swin-T, Params=88.0M2026.01—84.3489.11
SPBackbone=Visformer-T, Params=10.3M2026.01—82.1888.24
SP-CLIPBackbone=Visformer-T, Params=10.0M, FLOPs=1.3 × 10^92023.03—82.1888.24
SP-GloVeBackbone=Visformer-T, Params=10.0M, FLOPs=1.3 × 10^92023.03—81.6288.32
SP-SBERTBackbone=Visformer-T, Params=10.0M, FLOPs=1.3 × 10^92023.03—81.3288.31
SUNBackbone=Visformer-S, Params=12.4M, FLOPs=1.7 × 10^82023.03—78.3788.84
SVMSetting=Inductive, Backbone=ResNet-12, Input Size=84x842020.03—60.9479.93
SVM + ICISetting=Transductive, Backbone=ResNet-12, Input Size=84x842020.03—73.1683.72
SVM + ICISetting=Semi-supervised, Backbone=ResNet-12, Input Size=84x84, Unlabeled Samples=15/152020.03—72.5283.23
SVM + ICISetting=Semi-supervised, Backbone=ResNet-12, Input Size=84x84, Unlabeled Samples=30/502020.03—75.584
TEAMSetting=Transductive, Backbone=ResNet-18, Input Size=224x2242020.03—70.4381.25
TPNSetting=Transductive, Backbone=ResNet-12, Input Size=84x842020.03—65.8979.38