• Stars
    star
    367
  • Rank 116,257 (Top 3 %)
  • Language
    Python
  • Created over 4 years ago
  • Updated almost 4 years ago

Reviews

There are no reviews yet. Be the first to send feedback to the community and the maintainers!

Repository Details

RFS

Representations for Few-Shot Learning (RFS). This repo covers the implementation of the following paper:

"Rethinking few-shot image classification: a good embedding is all you need?" Paper, Project Page

If you find this repo useful for your research, please consider citing the paper

@article{tian2020rethink,
  title={Rethinking few-shot image classification: a good embedding is all you need?},
  author={Tian, Yonglong and Wang, Yue and Krishnan, Dilip and Tenenbaum, Joshua B and Isola, Phillip},
  journal={arXiv preprint arXiv:2003.11539},
  year={2020}
}

Installation

This repo was tested with Ubuntu 16.04.5 LTS, Python 3.5, PyTorch 0.4.0, and CUDA 9.0. However, it should be compatible with recent PyTorch versions >=0.4.0

Download Data

The data we used here is preprocessed by the repo of MetaOptNet, but we have renamed the file. Our version of data can be downloaded from here:

[DropBox]

Pre-trained Models

[DropBox]

Running

Exemplar commands for running the code can be found in scripts/run.sh.

For unuspervised learning methods CMC and MoCo, please refer to the CMC repo.

Contacts

For any questions, please contact:

Yonglong Tian ([email protected])
Yue Wang ([email protected])

Acknowlegements

Part of the code for distillation is from RepDistiller repo.