Few Shot Image Classification On Mini 3

Metriken

Accuracy

Ergebnisse

Leistungsergebnisse verschiedener Modelle zu diesem Benchmark

Modellname	Accuracy	Paper Title
pseudo-shots	82.51	Extended Few-Shot Learning: Exploiting Existing Resources for Novel Tasks
P>M>F (P=DINO-ViT-base, M=ProtoNet)	98.4	Pushing the Limits of Simple Pipelines for Few-Shot Learning: External Data and Fine-Tuning Make a Difference
EASY 2xResNet12 1/√2 (transductive)	88.57	EASY: Ensemble Augmented-Shot Y-shaped Learning: State-Of-The-Art Few-Shot Classification with Simple Ingredients
Illumination Augmentation	89.14	Sill-Net: Feature Augmentation with Separated Illumination Representation
EPNet + SSL	88.05	Embedding Propagation: Smoother Manifold for Few-Shot Classification
Adaptive Subspace Network	81.65	Adaptive Subspaces for Few-Shot Learning
TMDlayer + Transduction	77.78	Neural TMDlayer: Modeling Instantaneous flow of features via SDE Generators
ICI	80.11	Instance Credibility Inference for Few-Shot Learning
GAP	71.55	Meta-Learning with a Geometry-Adaptive Preconditioner
SemFew-Trans	86.49	Simple Semantic-Aided Few-Shot Learning
DiffKendall (Meta-Baseline, ResNet-12)	80.79	DiffKendall: A Novel Approach for Few-Shot Learning with Differentiable Kendall's Rank Correlation
COSOC (inductive)	85.16	Rectifying the Shortcut Learning of Background for Few-Shot Learning
HCTransformers	89.19	Attribute Surrogates Learning and Spectral Tokens Pooling in Transformers for Few-shot Learning
BD-CSPN + ESFR (WRN)	84.36	Unsupervised Embedding Adaptation via Early-Stage Feature Reconstruction for Few-Shot Classification
VAMPIRE	64.31	Uncertainty in Model-Agnostic Meta-Learning using Variational Inference
MAML	63.1	Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks
EASY 2xResNet12 1/√2 (inductive)	86.28	EASY: Ensemble Augmented-Shot Y-shaped Learning: State-Of-The-Art Few-Shot Classification with Simple Ingredients
PrototypeCompletion	84.18	Prototype Completion for Few-Shot Learning
RENet	82.58	Relational Embedding for Few-Shot Classification
LaplacianShot	84.72	Laplacian Regularized Few-Shot Learning

0 of 95 row(s) selected.