/fall-detection-kalman

A fall detection system using deep learning and Kalman filters to process signals from the Sysfall dataset. This project leverages advanced techniques to accurately detect falls, improving safety and response times in various environments.

Primary LanguageJupyter Notebook

Fall Detection System using Deep Learning and Kalman Filters

This project implements a fall detection system using deep learning techniques and Kalman filters to process signals from the Sysfall dataset. The system aims to accurately detect falls, enhancing safety and improving response times.

Repository Structure

  • /: Jupyter notebooks containing the implementation.
  • figs/: Directory containing the figures generated during the project.

Installation

  1. Clone the repository: 
    git clone https://github.com/Ilia-Abolhasani/fall-detection-kalman.git
cd fall-detection-kalman
  2. Install the required packages: 
        pip install python-bidi
    pip install arabic-reshaper
    pip install scipy

Project Overview

Signal Processing

Accelerometer Signals:

Before Kalman Filter:

Accelerator Sensor Before Kalman

After Kalman Filter:

Accelerator Sensor After Kalman

Gyroscope Signals

Before Kalman Filter:

Gyroscope Sensor Before Kalman

After Kalman Filter:

Gyroscope Sensor After Kalman

Signal Magnitude Area (SMA) Comparison

Accelerometer:

Accelerator SMA Compare

Gyroscope:

Gyroscope SMA Compare

Data Analysis

Activity Frequency

Activity Frequency

State Frequency

State Frequency

Distribution Standardization

Distribution Standardization

State Label Example

State Label Example

Model and Performance

Deep Learning Model Architecture

Deep Learning Model

State Confusion Matrix

State Confusion Matrix

Usage

Run the Jupyter notebooks to reproduce the results and visualizations shown above.

Contributing

Feel free to submit issues or pull requests if you have suggestions for improvements or find any bugs.

License

This project is licensed under the MIT License.