saidulk/gui_predictor_bl

Human Activity Recognition with smartphone sensors

In this project, A system was developed to collect smartphone IMU data in realtime and predict peformed activities from it.

Description

The system consists of 2 important parts:

Sensor data streaming and processing

Sensor data from smartphone was streamed in realtime using a custom made android application. The app used bluetooth to establish connection and transfer data to the PC. A python program was written using the pybluez library that maintained connection with the app and stored data in a buffer. The data was processed and fed to the deep-learning model to predict performed activity.

Graphical User Interface (GUI)

A Graphical User Interface (GUI) was built using python library PyQt5 for the sytem.

Flowchart

Dependencies

Data processing: numpy, matplotlib, pandas, pickle

GUI: PyQt5, pyqtgraph

Machine learning: sklearn, tensorflow, keras

Bluetooth: pybluez

Download and Usage

App link: https://github.com/saidulK/bluetooth_data_streamer

1. Download the app and pair smartphone with the PC.
2. Download this repo
3. Install dependencies
4. Run GUI_predictor.py
5. Open app