---
title: Hyperpose 紀錄
tags: Hyperpose, Openpose, OpenPifPaf, c++, tensorrt
description: Hyperpose 使用紀錄
---

# Hyperpose introduction
[**Hyperpose**](https://github.com/tensorlayer/hyperpose) is a tool to estimation human pose, and it is based on Openpose, OpenPifPaf. It provides a converted tensorrt model to accelerate inference speed.
Following [**Hyperpose document**](https://hyperpose.readthedocs.io/en/latest/markdown/quick_start/prediction.html#table-of-flags-for-hyperpose-cli) instruction to build the environment.

## Simple test result
provide a human pose estimation including 18 key point of a human.


## Use pose estimatnion if person fall down

Method 1: Generate the pose and use ST-GCN model to predict the action.
* https://github.com/GajuuzZ/Human-Falling-Detect-Tracks

Method 2: Generate the pose and use arglorithm to judge if person fall down.
* https://blog.csdn.net/weixin_26731327/article/details/108971459)
* https://towardsdatascience.com/fall-detection-using-pose-estimation-a8f7fd77081d

---
## 參考
* https://github.com/tensorlayer/hyperpose
* https://hyperpose.readthedocs.io/en/latest/markdown/quick_start/prediction.html#table-of-flags-for-hyperpose-cli
* https://blog.csdn.net/weixin_26731327/article/details/108971459
* https://github.com/GajuuzZ/Human-Falling-Detect-Tracks
* https://towardsdatascience.com/fall-detection-using-pose-estimation-a8f7fd77081d

## Thank you! :dash:

You can find me on
- GitHub: https://github.com/shaung08
- Email: [e-mail]