Improved Few-Shot Visual Classification

Peyman Bateni, Raghav Goyal, Vaden Masrani, Frank Wood, Leonid Sigal
12/7/2019
cs.CV

Official code: plai-group/simple-cnaps (58 stars, MIT, last commit 2022-11-03). 1 more implementation below.

Abstract

Few-shot learning is a fundamental task in computer vision that carries the promise of alleviating the need for exhaustively labeled data. Most few-shot learning approaches to date have focused on progressively more complex neural feature extractors and classifier adaptation strategies, as well as the refinement of the task definition itself. In this paper, we explore the hypothesis that a simple class-covariance-based distance metric, namely the Mahalanobis distance, adopted into a state of the art few-shot learning approach (CNAPS) can, in and of itself, lead to a significant performance improvement. We also discover that it is possible to learn adaptive feature extractors that allow useful estimation of the high dimensional feature covariances required by this metric from surprisingly few samples. The result of our work is a new "Simple CNAPS" architecture which has up to 9.2% fewer trainable parameters than CNAPS and performs up to 6.1% better than state of the art on the standard few-shot image classification benchmark dataset.

View on arXivView PDF

Code Implementations(2)

plai-group/simple-cnapsOfficial100%
MIT

Source codes for "Improved Few-Shot Visual Classification" (CVPR 2020), "Enhancing Few-Shot Image Classification with Unlabelled Examples" (WACV 2022), and "Beyond Simple Meta-Learning: Multi-Purpose Models for Multi-Domain, Active and Continual Few-Shot Learning" (Neural Networks 2022 - in submission)

5815Shell, PythonSep 18, 20213 years agoMIT
conditional-neural-processdeep-learningfew-shot-classifcationfew-shot-learningfew-shot-recognition+5 more
peymanbateni/simple-cnapsCommunity100%
MIT

Source codes for "Improved Few-Shot Visual Classification" (CVPR 2020), "Enhancing Few-Shot Image Classification with Unlabelled Examples" (WACV 2022), and "Beyond Simple Meta-Learning: Multi-Purpose Models for Multi-Domain, Active and Continual Few-Shot Learning" (SSRN Electronic Journal)

11925Shell, PythonNov 19, 20193 years agoMIT
conditional-neural-processdeep-learningfew-shot-classifcationfew-shot-learningfew-shot-recognition+5 more

Similar Papers

Diagnosing and Remedying Shot Sensitivity with Cosine Few-Shot Learners

88% match2022

Learning More Discriminative Local Descriptors for Few-shot Learning

87% match2023

Frequency Guidance Matters in Few-Shot Learning

87% match1 repo172023

Siamese Transformer Networks for Few-shot Image Classification

86% match2024

SuSana Distancia is all you need: Enforcing class separability in metric learning via two novel distance-based loss functions for few-shot image classification

86% match2023

A Closer Look at Few-shot Classification

85% match1 repo1.2k2019

Cite this paper

@article{bateni2019improved,
  title  = {Improved Few-Shot Visual Classification},
  author = {Peyman Bateni and Raghav Goyal and Vaden Masrani and Frank Wood and Leonid Sigal},
  year   = {2019},
  eprint = {1912.03432},
  archivePrefix = {arXiv},
  url    = {https://arxiv.org/abs/1912.03432}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/1912.03432.svg)](https://codeofpaper.com/paper/1912.03432)

Discussion