Skip to content

An official implementation of CVPR 2019 paper "All You Need Is a Few Shifts: Designing Efficient Convolutional Neural Networks for Image Classification".

License

Notifications You must be signed in to change notification settings

hikvision-research/SparseShiftLayer

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

All You Need Is a Few Shifts: Designing Efficient Convolutional Neural Networks for Image Classification

This repo is the official implementation of CVPR2019 paper "All You Need Is a Few Shifts: Designing Efficient Convolutional Neural Networks for Image Classification".

Paper introduction

SSL

The comparison of Sparse Shift Layer (SSL) with other shift operations applied to feature maps.

Preparation

Prerequistes

  • Python 3.6
  • Pytorch 1.0
  • CUDA 8.0 or higher

Usage

Data folds structure

.ImageNetOrigin/
  └── train/
  |      ├── n01440764
  |      |      ├── n01440764_18.JPEG
  |      |      ├── n01440764_36.JPEG
  |      |      └── ...
  |      ├── n01443537
  |      └── ...
  └── val2/
         ├── n01440764
         |      ├── n01440764_18.JPEG
         |      ├── n01440764_36.JPEG
         |      └── ...
         ├── n01443537
         └── ...

Train

bash train_imagenet.sh

Eval

bash test_imagenet.sh

Main Results

Networks MAdds Prarms Model Weights Top-1 TrainLog
FE-Net 1.0x 301M 3.7M To-Be-Released 73.6 % log-file
FE-Net 1.0x + Autoaugment 301M 3.7M To-Be-Released 73.9 % log-file
FE-Net 1.375x 563M 5.9M To-Be-Released 75.7 % log-file
FE-Net 1.375x + Autoaugment 563M 5.9M To-Be-Released 76.6 % log-file

Citation

Please consider citing our paper if you find it useful for your research.

@inproceedings{Chen2019ShiftNet,
  author = {Weijie Chen, Di Xie, Yuan Zhang, Shiliang Pu},
  title = {All You Need is a Few Shifts: Designing Efficient Convolutional Neural Networks for Image Classification},
  booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  year = {2019},
}

License

This project is released under the Apache 2.0 license. Other codes from open source repository follows the original distributive licenses.

Acknowledgement

We appreciate Xianfeng Li for transforming the source code from Caffe to Pytorch. If you have any problem about this work, please feel free to contact Xianfeng Li (lixianfeng6-at-hikvision.com) and Weijie Chen (chenweijie5-at-hikvision.com).

About

An official implementation of CVPR 2019 paper "All You Need Is a Few Shifts: Designing Efficient Convolutional Neural Networks for Image Classification".

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published