Few-Shot Classification Leaderboard

miniImageNet   tieredImageNet   Fewshot-CIFAR100   CIFAR-FS


   


The goal of this page is to keep on track of the state-of-the-arts (SOTA) for the few-shot classification.
Welcome to report results and revise mistakes by creating issues or pull requests.

miniImageNet Leaderboard (5-class)

Edit this leaderboard

Method Venue Year Backbone Setting 1-shot 5-shot Code Reported by
GNN + FT ICLR 2020 ResNet-10 Inductive 66.32 ± 0.80 81.98 ± 0.55 [PyTorch] [Source]
MetaOptNet CVPR 2019 ResNet-12 Inductive 62.64 ± 0.61 78.63 ± 0.46 [PyTorch] [Source]
MTL CVPR 2019 ResNet-12 Inductive 61.2 ± 1.8 75.5 ± 0.8 [PyTorch] [Source]
TADAM NeurIPS 2018 ResNet-12 Inductive 58.5 ± 0.3 76.7 ± 0.3 [TensorFlow] [Source]
MAML ICML 2017 4CONV Inductive 48.70 ± 1.75 63.11 ± 0.92 [TensorFlow] [Source]
LEO ICLR 2019 WRN-28-10 Inductive 61.76 ± 0.08 77.59 ± 0.12 [TensorFlow] [Source]
ProtoNets NeurIPS 2017 4CONV Inductive 49.42 ± 0.78 68.20 ± 0.66 [PyTorch] [Source]
MatchingNets NeurIPS 2016 4CONV Inductive 43.56 ± 0.84 55.31 ± 0.73 [TensorFlow] [Source]
RelationNets CVPR 2018 4CONV Inductive 50.44 ± 0.82 65.32 ± 0.70 [PyTorch] [Source]
TPN ICLR 2019 4CONV Transductive 55.51 ± 0.86 69.86 ± 0.65 [TensorFlow] [Source]
AdaResNet ICML 2018 ResNet-12 Inductive 56.88 ± 0.62 71.94 ± 0.57 None [Source]
DeepEMD CVPR 2020 ResNet-12 Inductive 65.91 ± 0.82 82.41 ± 0.56 [PyTorch] [Source]
ProtoNets NeurIPS 2017 ResNet-12 Inductive 60.37 ± 0.83 78.02 ± 0.57 [PyTorch] [Source]
MatchingNets NeurIPS 2016 ResNet-12 Inductive 63.08 ± 0.80 75.99 ± 0.60 [TensorFlow] [Source]
CTM CVPR 2019 ResNet-18 Inductive 64.12 ± 0.82 80.51 ± 0.13 [PyTorch] [Source]
wDAE-GNN CVPR 2019 WRN-28-10 Inductive 61.07 ± 0.15 76.75 ± 0.11 [PyTorch] [Source]
PPA CVPR 2018 WRN-28-10 Inductive 59.60 ± 0.41 73.74 ± 0.19 None [Source]
CAN NeurIPS 2019 ResNet-12 Inductive 63.85 ± 0.48 79.44 ± 0.34 [PyTorch] [Source]
CAN+T NeurIPS 2019 ResNet-12 Transductive 67.19 ± 0.55 80.64 ± 0.35 [PyTorch] [Source]
LGM-Net ICML 2019 6CONV Inductive 69.13 ± 0.35 71.18 ± 0.68 [TensorFlow] [Source]
FEAT CVPR 2020 ResNet-12 Inductive 66.78 ± 0.20 82.05 ± 0.14 [PyTorch] [Source]
FEAT CVPR 2020 4CONV Inductive 55.15 ± 0.20 71.61 ± 0.16 [PyTorch] [Source]
FEAT CVPR 2020 WRN-28-10 Inductive 65.10 ± 0.20 81.11 ± 0.14 [PyTorch] [Source]
Dhillon et al. ICLR 2020 WRN-28-10 Transductive 65.73 ± 0.68 78.40 ± 0.52 None [Source]
SIB ICLR 2020 WRN-28-10 Transductive 70.0 ± 0.6 79.2 ± 0.4 [PyTorch] [Source]
SIB ICLR 2020 4CONV Transductive 63.26 ± 1.07 75.73 ± 0.71 [PyTorch] [Source]
KTN ICCV 2019 4CONV Inductive 64.42 ± 0.72 74.16 ± 0.56 None [Source]
Ravichandran et al. ICCV 2019 4CONV Inductive 49.07 ± 0.43 65.73 ± 0.36 None [Source]
Ravichandran et al. ICCV 2019 ResNet-12 Inductive 60.71 77.64 None [Source]
Variational method ICCV 2019 ResNet-12 Inductive 61.23 ± 0.26 77.69 ± 0.17 None [Source]
Variational method ICCV 2019 4CONV Inductive 57.15 ± 0.31 71.54 ± 0.23 None [Source]
TEAM ICCV 2019 4CONV Transductive 56.57 72.04 None [Source]
TEAM ICCV 2019 ResNet-12 Transductive 60.07 75.90 None [Source]
Robust 20-dist++ ICCV 2019 ResNet-12 Inductive 63.73 ± 0.62 81.19 ± 0.43 [PyTorch] [Source]
DeepEMD v2 arXiv 2020 ResNet-12 Inductive 68.77 ± 0.29 84.13 ± 0.53 [PyTorch] [Source]
E3BM ECCV 2020 ResNet-12 Inductive 63.8 ± 0.4 80.1 ± 0.3 [PyTorch] [Source]
E3BM ECCV 2020 WRN-28-10 Transductive 71.4 ± 0.5 81.2 ± 0.4 [PyTorch] [Source]
LST NeurIPS 2019 ResNet-12 Semi-supervised 70.1 ± 1.9 78.7 ± 0.8 [TensorFlow] [Source]
AM3 NeurIPS 2019 ResNet-12 Inductive 65.30 ± 0.49 78.10 ± 0.36 [TensorFlow] [Source]
DSN-MR CVPR 2020 ResNet-12 Transductive 64.60 ± 0.72 79.51 ± 0.50 [PyTorch] [Source]
DPGN CVPR 2020 ResNet-12 Transductive 67.77 ± 0.32 84.60 ± 0.43 [PyTorch] [Source]
DPGN CVPR 2020 4CONV Transductive 66.01 ± 0.36 82.83 ± 0.41 [PyTorch] [Source]
AFHN CVPR 2020 ResNet-18 Inductive 62.38 ± 0.72 78.16 ± 0.56 None [Source]
LR+ICI CVPR 2020 ResNet-12 Transductive 66.80 79.26 [PyTorch] [Source]
LR+ICI CVPR 2020 ResNet-12 Semi-supervised 69.66 80.11 [PyTorch] [Source]
LR+ICI v2 arXiv 2020 ResNet-12 Semi-supervised 71.11 ± 1.15 81.25 ± 0.65 [PyTorch] [Source]
LR+ICI v2 arXiv 2020 ResNet-12 Transductive 68.70 ± 1.10 79.69 ± 0.66 [PyTorch] [Source]
TransMatch CVPR 2020 WRN-28-10 Semi-supervised 63.02 ± 1.07 82.24 ± 0.59 None [Source]
∆-encoder NeurIPS 2018 ResNet-18 Inductive 59.9 69.7 [TensorFlow] [Source]
Neg-Cosine ECCV 2020 ResNet-12 Inductive 63.85 ± 0.81 81.57 ± 0.56 [PyTorch] [Source]
J. Kim et al. ECCV 2020 ResNet-12 Inductive 65.08 ± 0.86 82.70 ± 0.54 [PyTorch] [Source]
BD-CSPN ECCV 2020 WRN-28-10 Transductive 70.31 ± 0.93 81.89 ± 0.60 None [Source]
Centroid alignment ECCV 2020 WRN-28-10 Inductive 65.92 ± 0.60 82.85 ± 0.55 [PyTorch] [Source]
ICA + MSP ECCV 2020 DenseNet Transductive 77.06 ± 0.26 84.99 ± 0.14 None [Source]
ICA + MSP ECCV 2020 DenseNet Semi-supervised 80.11 ± 0.25 85.78 ± 0.13 None [Source]
Y. Tian et al. ECCV 2020 ResNet-12 Inductive 64.82 ± 0.60 82.14 ± 0.43 [PyTorch] [Source]
EPNet ECCV 2020 WRN-28-10 Transductive 70.74 ± 0.85 84.34 ± 0.53 [PyTorch] [Source]
EPNet ECCV 2020 WRN-28-10 Semi-supervised 79.22 ± 0.92 88.05 ± 0.51 [PyTorch] [Source]
F. Wu et al. ECCV 2020 Capsule Network Inductive 66.43 ± 0.26 82.13 ± 0.21 None [Source]
Su et al. ECCV 2020 ResNet-18 Inductive - 76.6 ± 0.7 [PyTorch] [Source]
TIM NeurIPS 2020 ResNet-18 Transductive 73.9 85.0 [PyTorch] [Source]
TIM NeurIPS 2020 WRN-28-10 Transductive 77.8 87.4 [PyTorch] [Source]
IFSL NeurIPS 2020 WRN-28-10 Transductive 73.51 83.21 [PyTorch] [Source]
DKT NeurIPS 2020 4CONV Inductive 49.73 ± 0.07 - [PyTorch] [Source]
LR+DC ICLR 2021 WRN-28-10 Inductive 68.57 ± 0.55 82.88 ± 0.42 [PyTorch] [Source]
IEPT ICLR 2021 ResNet-12 Inductive 67.05 ± 0.44 82.90 ± 0.30 None [Source]
IEPT ICLR 2021 4CONV Inductive 56.26 ± 0.45 73.91 ± 0.34 None [Source]
MELR ICLR 2021 ResNet-12 Inductive 67.40 ± 0.43 83.40 ± 0.28 None [Source]
MELR ICLR 2021 4CONV Inductive 55.35 ± 0.43 72.27 ± 0.35 None [Source]
ConstellationNet ICLR 2021 ResNet-12 Inductive 64.89 ± 0.23 79.95 ± 0.37 None [Source]
OVE PG G P+ Cosine (ML) ICLR 2021 4CONV Inductive 50.02 ± 0.35 64.58 ± 0.31 None [Source]
OVE PG G P+ Cosine (PL) ICLR 2021 4CONV Inductive 48.00 ± 0.24 67.14 ± 0.23 None [Source]
BOIL ICLR 2021 4CONV Inductive 49.61 ± 0.16 66.45 ± 0.37 None [Source]
FRN CVPR 2021 ResNet-12 Inductive 66.45 ± 0.19 82.83 ± 0.13 [PyTorch] [Source]
PT+MAP arXiv 2021 WRN Inductive 65.35 ± 0.20 83.87 ± 0.13 [PyTorch] [Source]
PT+MAP arXiv 2021 WRN Transductive 82.92 ± 0.26 88.82 ± 0.13 [PyTorch] [Source]
CSEI AAAI 2021 ResNet-12 Inductive 68.94 ± 0.28 85.07 ± 0.50 None [Source]
PTN AAAI 2021 WRN-28-10 Semi-supervised 82.66 ± 0.97 88.43 ± 0.67 None [Source]
P-Transfer AAAI 2021 ResNet-12 Inductive 64.21 ± 0.77 80.38 ± 0.59 None [Source]
CNL AAAI 2021 ResNet-12 Inductive 67.96 ± 0.98 83.36 ± 0.51 None [Source]
TACO AAAI 2021 ResNet-12 Semi-supervised 68.23 83.42 None [Source]

Contact

Yaoyao Liu / yaoyao.liu (at) mpi-inf.mpg.de

Chi Zhang / chi007 (at) e.ntu.edu.sg