Few Shot Image Classification On Meta Dataset

评估指标

Accuracy

评测结果

各个模型在此基准测试上的表现结果

模型名称	Accuracy	Paper Title	Repository
Transductive CNAPS	70.32	Enhancing Few-Shot Image Classification with Unlabelled Examples
Prototypical Networks	60.573	Prototypical Networks for Few-shot Learning
Matching Networks	56.247	Matching Networks for One Shot Learning
Simple CNAPS	69.86	Improved Few-Shot Visual Classification
TSP (ResNet18; applied on TA^2-Net)	81.40	Task-Specific Preconditioner for Cross-Domain Few-Shot Learning	-
URT+MQDA	74.3	Shallow Bayesian Meta Learning for Real-World Few-Shot Recognition
fo-Proto-MAML	63.428	Meta-Dataset: A Dataset of Datasets for Learning to Learn from Few Examples
UpperCaSE-EfficientNetB0	76.1	Contextual Squeeze-and-Excitation for Efficient Few-Shot Image Classification
TSA (ResNet18, URL, residual adapters, 84x84 image, shuffled data, scratch, MDL)	78.07	Cross-domain Few-shot Learning with Task-specific Adapters
k-NN	54.319	Meta-Dataset: A Dataset of Datasets for Learning to Learn from Few Examples
SUR	70.72	Selecting Relevant Features from a Multi-domain Representation for Few-shot Classification
SUR-pnf	69.3	Selecting Relevant Features from a Multi-domain Representation for Few-shot Classification
URL (ResNet18, 84x84 image, shuffled data, scratch, MDL)	75.75	Universal Representation Learning from Multiple Domains for Few-shot Classification
Invariance-Equivariance	68.89	Exploring Complementary Strengths of Invariant and Equivariant Representations for Few-Shot Learning
Finetune	58.758	Meta-Dataset: A Dataset of Datasets for Learning to Learn from Few Examples
UpperCaSE-ResNet50	74.9	Contextual Squeeze-and-Excitation for Efficient Few-Shot Image Classification
fo-MAML	57.024	Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
URT	72.15	A Universal Representation Transformer Layer for Few-Shot Image Classification
CNAPs	66.9	Fast and Flexible Multi-Task Classification Using Conditional Neural Adaptive Processes
P>M>F (P=DINO-ViT-base, M=ProtoNet)	84.75	Pushing the Limits of Simple Pipelines for Few-Shot Learning: External Data and Fine-Tuning Make a Difference

0 of 22 row(s) selected.