snoymy / AI_fall_detection

AI Fall Detection - How to run

Content

• Requirement

  • Software

  • Library

    • Installation

  • Dataset

  • Source Code

  • Model

  • Clone from github

• Folder Structure

• Run

  • Train and predict

  • Predict only

Requirement

Software

• Anaconda

• Spyder

Library

• MediaPipe

• TensorFlow

Installation

Anaconda

If you are windows user follow steps from this link.

MediaPipe

pip install mediapipe

Tensorflow

pip install tensorflow

Dataset

Please download from this link.

Source Code

Please download main.py , predict.py and videorecorder.py from this link.

Model

Please download from this link (Lastest version is dense4-2x1000-40vid.h5).

Clone from github

via Terminal

git clone https://github.com/snoymy/AI_fall_detection.git

via web link

Folder Structure

After install libray , dataset and source code , Please create project root directory Ex. fall_detector and place all file you had download in the folder

fall_detector
  ├── train_video
  │    ├── fall
  │    └── normal
  ├── main.py
  ├── predict.py
  ├── videorecorder.py
  └── dense4-2x1000-40vid.h5

Run

Train and predict

Open main.py with Spyder and press run. This will collect data -> train -> predict.

after run

and predict

you can create new dataset by your self by runningvideorecorder.py.

Predict only

First you need to have model (can download from Requirement section). Open file predict.py and run.