/ghostnet.pytorch

[CVPR2020] Surpassing MobileNetV3: "GhostNet: More Features from Cheap Operations"

Primary LanguagePython

GhostNet

This repo provides Pytorch implementation of GhostNet: More Features from Cheap Operations. The TensorFlow implementation with pretrained model is available at here.

Requirements

The code was verified on Python3, PyTorch 1.0+.

Usage

Usage example:

import torch
from ghost_net import ghost_net

model = ghost_net(width_mult=1.0)
input = torch.randn(32,3,224,224)
y = model(input)
print(y)

Introduction of GhostNet

GhostNet: More Features from Cheap Operations [arXiv]

By Kai Han, Yunhe Wang, Qi Tian, Jianyuan Guo, Chunjing Xu, Chang Xu.

  • Approach
  • Performance

We beat other SOTA lightweight CNNs such as MobileNetV3 and FBNet.

Citation

@article{ghostnet,
  title={GhostNet: More Features from Cheap Operations},
  author={Han, Kai and Wang, Yunhe and Tian, Qi and Guo, Jianyuan and Xu, Chunjing and Xu, Chang},
  journal={arXiv},
  year={2019}
}