Awesome
<div align="center"> <h2><a href="https://arxiv.org/abs/2103.13027">AutoMix: Unveiling the Power of Mixup for Stronger Classifiers</a></h2> (ECCV 2022 Oral)Zicheng Liu<sup>*,1,2</sup>, Siyuan Li<sup>*,1,2</sup>, Di Wu<sup>1,2</sup>, Zhiyuan Chen<sup>1</sup>, Lirong Wu<sup>1,2</sup>, Stan Z. Li<sup>†,1</sup>
<sup>1</sup>Westlake University, <sup>2</sup>Zhejiang University
</div> <p align="center"> <a href="https://arxiv.org/abs/2103.13027" alt="arXiv"> <img src="https://img.shields.io/badge/arXiv-2210.13452-b31b1b.svg?style=flat" /></a> <a href="https://github.com/Westlake-AI/AutoMix/blob/main/LICENSE" alt="license"> <img src="https://img.shields.io/badge/license-Apache--2.0-%23B7A800" /></a> <!-- <a href="https://colab.research.google.com/github/Westlake-AI/MogaNet/blob/main/demo.ipynb" alt="Colab"> <img src="https://colab.research.google.com/assets/colab-badge.svg" /></a> --> <a href="https://zhuanlan.zhihu.com/p/550300558" alt="license"> <img src="https://img.shields.io/badge/zhihu-automix-blue" /></a> </p>We propose a novel automatic mixup (AutoMix) framework, where the mixup policy is parameterized and serves the ultimate classification goal directly. Specifically, AutoMix reformulates the mixup classification into two sub-tasks (i.e., mixed sample generation and mixup classification) with corresponding sub-networks and solves them in a bi-level optimization framework. For the generation, a learnable lightweight mixup generator, Mix Block, is designed to generate mixed samples by modeling patch-wise relationships under the direct supervision of the corresponding mixed labels. To prevent the degradation and instability of bi-level optimization, we further introduce a momentum pipeline to train AutoMix in an end-to-end manner. Extensive experiments on nine image benchmarks prove the superiority of AutoMix compared with state-of-the-arts in various classification scenarios and downstream tasks.
<p align="center"> <img src="https://user-images.githubusercontent.com/44519745/174272662-19ce57ad-7b08-4e73-81b1-3bb81fee2fe5.png" width=100% height=100% class="center"> </p> <!-- <details> <summary>Table of Contents</summary> <ol> <li><a href="#catalog">Catalog</a></li> <li><a href="#image-classification">Image Classification</a></li> <li><a href="#license">License</a></li> <li><a href="#acknowledgement">Acknowledgement</a></li> <li><a href="#citation">Citation</a></li> </ol> </details> -->Catalog
We plan to update this timm implementation of AutoMix in a few months. Please watch us for the latest release or use our OpenMixup implementations.
- Image Classification Code with OpenMixup [code]
- CIFAR-10/100 and Tiny-ImageNet Training and Validation Code with timm [code]
- ImageNet-1K Training and Validation Code [code]
- Image Classification on Google Colab and Notebook Demo
Installation
Please check INSTALL.md for installation instructions.
Small-scale Image Classification
Please refer to OpenMixup implementations of CIFAR-100 and Tiny-ImageNet.
ImageNet Classification
1. Training and Validation
See TRAINING.md for ImageNet-1K training and validation instructions, or refer to our OpenMixup implementations. We released pre-trained models on OpenMixup.
<!-- Here is a notebook [demo](demo.ipynb) of AutoMix which run the steps to perform inference for image classification and generate mixup samples. -->2. ImageNet-1K Trained Models
Please refer to mixup_benchmarks in OpenMixup implementations for results and models.
<p align="right">(<a href="#top">back to top</a>)</p>License
This project is released under the Apache 2.0 license.
Acknowledgement
Our implementation is mainly based on the following codebases. We gratefully thank the authors for their wonderful works.
- pytorch-image-models: PyTorch image models, scripts, pretrained weights.
- OpenMixup: CAIRI Supervised, Semi- and Self-Supervised Visual Representation Learning Toolbox and Benchmark.
Citation
If you find this repository helpful, please consider citing:
@InProceedings{liu2022automix,
title={AutoMix: Unveiling the Power of Mixup for Stronger Classifiers},
author={Zicheng Liu and Siyuan Li and Di Wu and Zhiyuan Chen and Lirong Wu and Jianzhu Guo and Stan Z. Li},
booktitle={European Conference on Computer Vision},
pages={441--458},
year={2022},
}
<p align="right">(<a href="#top">back to top</a>)</p>