In this project, A system was developed to collect smartphone IMU data in realtime and predict peformed activities from it.
The system consists of 2 important parts:
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.
A Graphical User Interface (GUI) was built using python library PyQt5 for the sytem.
Data processing: numpy, matplotlib, pandas, pickle
GUI: PyQt5, pyqtgraph
Machine learning: sklearn, tensorflow, keras
Bluetooth: pybluez
App link: https://github.com/saidulK/bluetooth_data_streamer
- Download the app and pair smartphone with the PC.
- Download this repo
- Install dependencies
- Run GUI_predictor.py
- Open app