# Halpe-FullBody

**Repository Path**: james9/Halpe-FullBody

## Basic Information

- **Project Name**: Halpe-FullBody
- **Description**: Halpe: full body human pose estimation and human-object interaction detection dataset
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2021-10-27
- **Last Updated**: 2021-10-27

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# Halpe Full-Body Human Keypoints and HOI-Det dataset

## What is Halpe?
**Halpe** is a joint project under [AlphaPose](https://github.com/MVIG-SJTU/AlphaPose) and [HAKE](http://hake-mvig.cn/).  It aims at pushing <b>Human Understanding</b> to the extreme. We provide detailed annotation of human keypoints, together with the human-object interaction trplets from HICO-DET. For each person, we annotate 136 keypoints in total, including head,face,body,hand and foot. Below we provide some samples of Halpe dataset.
Halpe 是 AlphaPose 和 HAKE 旗下的一个联合项目。 它旨在将人类理解推向极致。 我们提供了人类关键点的详细注释，以及来自 HICO-DET 的人-物交互 trplets。 对于每个人，我们总共注释了 136 个关键点，包括头、脸、身体、手和脚。 下面我们提供一些 Halpe 数据集的样本。 
<div align="center">
    <img src="docs/1.jpg", width="400" alt>
    <img src="docs/2.jpg", width="400" alt>
    <img src="docs/4.jpg", width="250" alt>
    <img src="docs/5.jpg", width="250" alt>
    <img src="docs/3.jpg", width="300" alt>
</div>

## Download
Train annotations [[Baidu](https://pan.baidu.com/s/1hWX-I-HpXXnLcprFskfriQ) | [Google](https://drive.google.com/file/d/13vj8H0GZ9yugLPhVVWV9fcH-3RyW5Wk5/view?usp=sharing) ]

Val annotations [[Baidu](https://pan.baidu.com/s/1yDvBkXTSwk20EjiYzimpPw) | [Google](https://drive.google.com/file/d/1FdyCgro2t9_nOhTlMPjEf3c0aLOz9wi6/view?usp=sharing) ]

Train images from [HICO-DET](https://drive.google.com/open?id=1QZcJmGVlF9f4h-XLWe9Gkmnmj2z1gSnk)

Val images from [COCO](http://images.cocodataset.org/zips/val2017.zip)

## Realtime Demo with tracking（带跟踪的实时演示 ）
Trained model is available in [AlphaPose](https://github.com/MVIG-SJTU/AlphaPose)! 
训练模型在 [AlphaPose] 中可用 
Check out its [MODEL_ZOO](https://github.com/MVIG-SJTU/AlphaPose/blob/master/docs/MODEL_ZOO.md)
看看它的[MODEL_ZOO]
<div align="center">
    <img src="docs/alphapose_136_2.gif", width="400" alt>
</div>


## Keypoints format(关键点格式 )
We annotate 136 keypoints in total(我们总共注释了 136 个关键点： ):
```
    //26 body keypoints
    {0,  "Nose"},
    {1,  "LEye"},
    {2,  "REye"},
    {3,  "LEar"},
    {4,  "REar"},
    {5,  "LShoulder"},
    {6,  "RShoulder"},
    {7,  "LElbow"},
    {8,  "RElbow"},
    {9,  "LWrist"},
    {10, "RWrist"},
    {11, "LHip"},
    {12, "RHip"},
    {13, "LKnee"},
    {14, "Rknee"},
    {15, "LAnkle"},
    {16, "RAnkle"},
    {17,  "Head"},
    {18,  "Neck"},
    {19,  "Hip"},
    {20, "LBigToe"},
    {21, "RBigToe"},
    {22, "LSmallToe"},
    {23, "RSmallToe"},
    {24, "LHeel"},
    {25, "RHeel"},
    //face
    {26-93, 68 Face Keypoints}
    //left hand
    {94-114, 21 Left Hand Keypoints}
    //right hand
    {115-135, 21 Right Hand Keypoints}
## 中文
{0, "鼻子"},
     {1, "L 眼"},
     {2, "右眼"},
     {3, "L 耳朵"},
     {4, "右耳"},
     {5, "L 肩"},
     {6, "R 肩"},
     {7, "L 肘"},
     {8, "R 弯头"},
     {9, "L 手腕"},
     {10, "右手腕"},
     {11, "L 臀"},
     {12, "右臀"},
     {13, "LKnee"},
     {14, "右膝"},
     {15, "L 脚踝"},
     {16, "右脚踝"},
     {17, "头"},
     {18, "脖子"},
     {19, "臀部"},
     {20, "大脚趾"},
     {21, "R 大脚趾"},
     {22, "L 小脚趾"},
     {23, "R 小脚趾"},
     {24, "L 跟"},
     {25, "R 跟"},
     //脸
     {26-93, 68 人脸关键点}
     //左手
     {94-114, 21 左手关键点}
     //右手
     {115-135, 21 右手关键点} 
```

Illustration(插图):
<div align="center">
    <img src="docs/human_model.jpg", width="400" alt><br>
    26 body keypoints
</div>
<div align="center">
    <img src="docs/face.jpg", width="400" alt><br>
    68 face keypoints
</div>
<div align="center">
    <img src="docs/hand.jpg", width="400" alt><br>
    21 hand keypoints
</div>

## Usage(用法 )
The annotation is in the same format as COCO dataset. For usage, a good start is to check out the `vis.py`. We also provide related APIs. See [halpecocotools](https://github.com/HaoyiZhu/HalpeCOCOAPI), which can be installed by `pip install halpecocotools`.
注释与 COCO 数据集的格式相同。 对于使用，一个好的开始是查看`vis.py`。 我们还提供相关的 API。 参见[halpecocotools](https://github.com/HaoyiZhu/HalpeCOCOAPI)，可以通过`pip install halpecocotools`安装。 
## Evaluation(评估)
We adopt the same evaluation metrics as COCO dataset. 
我们采用与 COCO 数据集相同的评估指标。
## Related resources(  相关资源 )
A concurrent work [COCO-WholeBody](https://github.com/jin-s13/COCO-WholeBody) also annotate the full body keypoints. And HOI-DET for COCO dataset is also available at [V-COCO](https://github.com/s-gupta/v-coco)
一项并发工作 [COCO-WholeBody](https://github.com/jin-s13/COCO-WholeBody) 也对全身关键点进行了注释。 COCO 数据集的 HOI-DET 也可在 [V-COCO](https://github.com/s-gupta/v-coco) 获得 
## Citation(引文 )
The paper introducing this project is coming soon.
If the data helps your research, please cite the following papers at present:
介绍该项目的论文即将发布。
如果数据对您的研究有帮助，请参考目前以下论文： 
```
@inproceedings{fang2017rmpe,
  title={{RMPE}: Regional Multi-person Pose Estimation},
  author={Fang, Hao-Shu and Xie, Shuqin and Tai, Yu-Wing and Lu, Cewu},
  booktitle={ICCV},
  year={2017}
}
@inproceedings{li2020pastanet,
  title={PaStaNet: Toward Human Activity Knowledge Engine},
  author={Li, Yong-Lu and Xu, Liang and Liu, Xinpeng and Huang, Xijie and Xu, Yue and Wang, Shiyi and Fang, Hao-Shu and Ma, Ze and Chen, Mingyang and Lu, Cewu},
  booktitle={CVPR},
  year={2020}
}
```