This repository is an official implementation of the paper "FYI: Flip Your Images for Dataset Distillation" using PyTorch.
For detailed information, please refer to our project site or paper.
Please install packages in the requirements.
- Prepare the dataset in
/dataset
directory for experiments onTiny-ImageNet
andImageNet subset
. - Run the provided scripts as below. Modify the scripts for different settings.
cd DC_DSA_DM
# DM for 10 IPC on CIFAR-10
./scripts/dm.sh
# DSA for 10 IPC on CIFAR-10
./scripts/dsa.sh
cd MTT
# Generating teacher trajectories on CIFAR-10 using ZCA whitening
./scripts/buffer.sh
# MTT for 1 IPC on CIFAR-10
./scripts/distill.sh
This implementation is built on DC/DSA/DM and MTT. We also applied our method to IDC and FTD in our paper. We thank the authors and contributors of these projects.
If you find our work useful for your research, please cite our paper:
@inproceedings{son2024fyi,
title={{FYI}: Flip Your Images for Dataset Distillation},
author={Son, Byunggwan and Oh, Youngmin and Baek, Donghyeon and Ham, Bumsub},
booktitle={Proceedings of the European Conference on Computer Vision (ECCV)},
year={2024}
}