# CV-Backbones

**Repository Path**: luyvlei/CV-Backbones

## Basic Information

- **Project Name**: CV-Backbones
- **Description**: No description available
- **Primary Language**: Python
- **License**: Apache-2.0
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2021-06-15
- **Last Updated**: 2021-06-15

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# CV Backbones (including GhostNet, TinyNet, etc.)
- [GhostNet Code](#ghostnet-code)
- [TinyNet Code](#tinynet-code)
- [Model Desription](#model-desription)
- [Citation](#citation)
- [Other versions](#other-versions)

**News** 

2020/11/10 The code of TinyNet (NeurIPS 2020) has been release at [MindSpore Model Zoo](https://gitee.com/mindspore/mindspore/tree/master/model_zoo/research/cv/tinynet).

2020/10/31 GhostNet+TinyNet achieves better performance. See details in our NeurIPS 2020 paper: [arXiv](https://arxiv.org/abs/2010.14819).

2020/09/24 We release GhostNet models for more vision tasks on [MindSpore Hub](https://www.mindspore.cn/resources/hub) and [MindSpore Model Zoo](https://gitee.com/mindspore/mindspore/tree/master/model_zoo/research/cv).

2020/06/10 GhostNet is included in [PyTorch Hub](https://pytorch.org/hub/pytorch_vision_ghostnet/).

---

## GhostNet Code

This repo provides **pretrained models** and **inference code** for TensorFlow and PyTorch:
- Tensorflow: [./ghostnet_tensorflow](https://github.com/huawei-noah/CV-backbones/tree/master/ghostnet_tensorflow) with pretrained model.
- PyTorch: [./ghostnet_pytorch](https://github.com/huawei-noah/CV-backbones/tree/master/ghostnet_pytorch) with pretrained model.
- We also opensource code on [MindSpore Hub](https://www.mindspore.cn/resources/hub) and [MindSpore Model Zoo](https://gitee.com/mindspore/mindspore/tree/master/model_zoo/research/cv).

For **training**, please refer to [tinynet](https://gitee.com/mindspore/mindspore/tree/master/model_zoo/research/cv/tinynet) or [timm](https://rwightman.github.io/pytorch-image-models/training_hparam_examples/#mobilenetv3-large-100-75766-top-1-92542-top-5).

## TinyNet Code

This repo provides **pretrained models** and **inference code** for PyTorch:
- PyTorch: [./tinynet_pytorch](https://github.com/huawei-noah/CV-backbones/tree/master/tinynet_pytorch) with pretrained model.
- We also opensource training code on [MindSpore Model Zoo](https://gitee.com/mindspore/mindspore/tree/master/model_zoo/research/cv).


## Model Desription

GhostNet: More Features from Cheap Operations. CVPR 2020. [[Paper]](http://openaccess.thecvf.com/content_CVPR_2020/html/Han_GhostNet_More_Features_From_Cheap_Operations_CVPR_2020_paper.html) [[arXiv]](https://arxiv.org/abs/1911.11907)

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

<div align="center">
   <img src="./fig/ghost_module.png" width="720">
</div>

GhostNet beats other SOTA lightweight CNNs such as MobileNetV3 and FBNet.

<div align="center">
   <img src="./fig/flops_latency.png" width="720">
</div>


## Citation
```
@inproceedings{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},
  booktitle={CVPR},
  year={2020}
}
@article{tinynet,
  title={Model Rubik’s Cube: Twisting Resolution, Depth and Width for TinyNets},
  author={Han, Kai and Wang, Yunhe and Zhang, Qiulin and Zhang, Wei and Xu, Chunjing and Zhang, Tong},
  journal={Advances in Neural Information Processing Systems},
  volume={33},
  year={2020}
}
```

## Other versions
This repo provides the TensorFlow/PyTorch code of GhostNet. Other versions and applications can be found in the following:

0. timm: [code with pretrained model](https://github.com/rwightman/pytorch-image-models/blob/master/timm/models/ghostnet.py)
1. Darknet: [cfg file](https://github.com/AlexeyAB/darknet/files/3997987/ghostnet.cfg.txt), and [description](https://github.com/AlexeyAB/darknet/issues/4418)
2. Gluon/Keras/Chainer: [code](https://github.com/osmr/imgclsmob)
3. Paddle: [code](https://github.com/PaddlePaddle/PaddleClas/blob/master/ppcls/modeling/architectures/ghostnet.py)
4. Bolt inference framework: [benckmark](https://github.com/huawei-noah/bolt/blob/master/docs/BENCHMARK.md)
5. Human pose estimation: [code](https://github.com/tensorboy/centerpose/blob/master/lib/models/backbones/ghost_net.py)
6. YOLO with GhostNet backbone: [code](https://github.com/HaloTrouvaille/YOLO-Multi-Backbones-Attention)
7. Face recognition: [cavaface](https://github.com/cavalleria/cavaface.pytorch/blob/master/backbone/ghostnet.py), [FaceX-Zoo](https://github.com/JDAI-CV/FaceX-Zoo)