# MobilePose

**Repository Path**: giteebob/MobilePose

## Basic Information

- **Project Name**: MobilePose
- **Description**: Light-weight Single Person Pose Estimator
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2022-01-26
- **Last Updated**: 2022-01-26

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# MobilePose

MobilePose is a **Tiny** PyTorch implementation of single person 2D pose estimation framework. The aim is to provide the interface of the training/inference/evaluation, and the dataloader with various data augmentation options. And final trained model can satisfy basic requirements(speed+size+accuracy) for mobile device.

Some codes for networks and display are brought from:
1. [pytorch-mobilenet-v2](https://github.com/tonylins/pytorch-mobilenet-v2)
2. [Vanilla FCN, U-Net, SegNet, PSPNet, GCN, DUC](https://github.com/zijundeng/pytorch-semantic-segmentation)
3. [Shufflenet-v2-Pytorch](https://github.com/ericsun99/Shufflenet-v2-Pytorch)
4. [tf-pose-estimation](https://github.com/ildoonet/tf-pose-estimation) 
5. [dsntnn](https://github.com/anibali/dsntnn)

## NEWS!

- Apr 2021: [Siyuan Pan](https://github.com/pansiyuan123) provides [MNN version](https://market.mnn.zone/s/#/modelmarket/detail/107)!
- Mar 2019: Support running on MacBook with decent FPS!  
- Feb 2019: **ALL** the pretrained model files are avaliable!

## Requirements

- Python 3.7
- PyTorch 1.0
- [dsntnn 1.0](https://github.com/anibali/dsntnn)

## Evaluation Results

|Model(+DUC+DSNTNN)|Parmas(M)|Flops(G)|AP@0.5:0.95|AP@0.5|AR@0.5:0.95|AR@0.5|Link|
|:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:|
|ResNet18|12.26|1.64|**68.2**|93.9|**79.7**|96.7|[51.5M](https://drive.google.com/open?id=17Z1zetIVDI4_8-ZoFgTRsjHtDpwGtjRT)|
|MobileNetV2|3.91|0.49|67.5|**94.9**|79.4|**97.1**|[16.6M](https://drive.google.com/open?id=15Ihv1bVQv6_tYTFlECJMNrXEmrrka5g4)|
|ShuffleNetV2|2.92|**0.31**|61.5|91.6|74.8|95.5|[12.4M](https://drive.google.com/open?id=184Zg4E6HbbizPFYcELMXCd7mwWXdUd3U)|
|SqueezeNet1.1|**2.22**|0.63|58.4|92.1|72.3|95.8|[9.3M](https://drive.google.com/open?id=1RePeiBJHeHvmYTQ5vAUJHC5CstHIBcP0)|

<div align="center">
    <img src="./demo.png">
</div>

## Features

- [x] multi-thread dataloader with augmentations (dataloader.py)
- [x] training and inference (training.py)
- [x] performance evaluation (eval.py)
- [x] multiple models support (network.py)
- [x] ipython notebook visualization (demo.ipynb)
- [x] Macbook camera realtime display script (run_webcam.py)

## Usage

1. Installation:

```shell
pip install -r requirements.txt
```
2. Training:
```shell
python training.py --model shufflenetv2 --gpu 0 --inputsize 224 --lr 1e-3 --batchsize 128 --t7 ./models/shufflenetv2_224_adam_best.t7
```
3. Evaluation
```shell
ln -s cocoapi/PythonAPI/pycocotools
cd cocoapi/PythonAPI && make

python eval.py --t7 ./models/resnet18_224_adam_best.t7 --model resnet18 --gpu 0
```
4. Web Camera Demo (MacBook)

```shell
python run_webcam.py --model squeezenet --inp_dim 224 --camera 0
```

## Contributors

MobilePose is developed and maintained by [Yuliang Xiu](http://xiuyuliang.cn/about/), [Zexin Chen](https://github.com/ZexinChen) and [Yinghong Fang](https://github.com/Fangyh09). Thanks for [Siyuan Pan](https://github.com/pansiyuan123)'s implementation of [mnn version](https://market.mnn.zone/s/#/modelmarket/detail/107).

## License

MobilePose is freely available for free non-commercial use. For commercial queries, please contact [Cewu Lu](http://www.mvig.org/).