# Impact of base dataset design on few-shot image classification

arXiv: https://arxiv.org/abs/2007.08872 · HTML: https://mlsanity.com/paper/2007.08872/

## Key facts

- Cited by 9 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 17 Jul 2020.

## Built on by (most cited first, 9 of 9)

- [A Closer Look at Few-shot Classification Again](https://mlsanity.com/paper/2301.12246/) (2023)
- [Mask-guided Vision Transformer (MG-ViT) for Few-Shot Learning](https://mlsanity.com/paper/2205.09995/) (2022)
- [Adaptive Dimension Reduction and Variational Inference for Transductive Few-Shot Classification](https://mlsanity.com/paper/2209.08527/) (2022)
- [Few-Shot Video Object Detection](https://mlsanity.com/paper/2104.14805/) (2021)
- [Few-shot Learning as Cluster-induced Voronoi Diagrams: A Geometric Approach](https://mlsanity.com/paper/2202.02471/) (2022)
- [Image Augmentation for Multitask Few-Shot Learning: Agricultural Domain Use-Case](https://mlsanity.com/paper/2102.12295/) (2021)
- [Understanding the Cross-Domain Capabilities of Video-Based Few-Shot Action Recognition Models](https://mlsanity.com/paper/2406.01073/) (2024)
- [Few-Shot Electronic Health Record Coding through Graph Contrastive Learning](https://mlsanity.com/paper/2106.15467/) (2021)
- [P3DC-Shot: Prior-Driven Discrete Data Calibration for Nearest-Neighbor Few-Shot Classification](https://mlsanity.com/paper/2301.00740/) (2023)

## Abstract (arXiv)

The quality and generality of deep image features is crucially determined by the data they have been trained on, but little is known about this often overlooked effect. In this paper, we systematically study the effect of variations in the training data by evaluating deep features trained on different image sets in a few-shot classification setting. The experimental protocol we define allows to explore key practical questions. What is the influence of the similarity between base and test classes? Given a fixed annotation budget, what is the optimal trade-off between the number of images per class and the number of classes? Given a fixed dataset, can features be improved by splitting or combining different classes? Should simple or diverse classes be annotated? In a wide range of experiments, we provide clear answers to these questions on the miniImageNet, ImageNet and CUB-200 benchmarks. We also show how the base dataset design can improve performance in few-shot classification more drastically than replacing a simple baseline by an advanced state of the art algorithm.