Skip to content

A video based fall detection approach and a fall detection system on ROS based on single image.

License

Notifications You must be signed in to change notification settings

RichardChen20/PoseFall

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

20 Commits
 
 
 
 
 
 
 
 

Repository files navigation

Posefall

This repository contains two fall detection projects.

The first one is a video-based fall detection approach https://arxiv.org/abs/2107.14633, which takes 300 frames of 2D poses to detect whether there is a fall behaviour. The experiments were conducted on NTU RGBD dataset, we refer to codes from https://github.com/kchengiva/DecoupleGCN-DropGraph to train and evaluate our model. Only codes for our fall detection model is provided in this repository because...I'm lazy... Cleaning up the codes is quite time-comsuming and my research interest is not action recognition.

2023.2 new! I have uploaded codes for video-based fall detection, however, I do not clean them, hope they would help. https://drive.google.com/drive/folders/1wQz-TSld67ZZljYgCtfUKhjscvpAMOr2?usp=sharing

The other is a fall detection system on ROS based on single frame 2D pose. Here is the demo: https://www.bilibili.com/video/BV1AT4y1u7t3/ Pretrained weights can be found here: https://drive.google.com/drive/folders/1s1hcetDzHP6DlVecV3Zj199TVsDVsjLI?usp=sharing

I do not plan to update this repository detaily, you can send messages to me if you are interested in this work.

Tips for "ROS_fall_detection": Remember to modify the first line in "cam.py" and "detector.py": #! /home/seanchen/anaconda3/bin/python. Change it to your own Python3 path as our rep uses pytorch with Python3, while original ros is based on Python2.

About

A video based fall detection approach and a fall detection system on ROS based on single image.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published