CollectAccelerometerDatav2

This project is designed to collect accelerometer data from a mobile device using an application, which is then sent to a computer for action prediction based on predefined actions.

Application Repository

The application used for data collection can be found at CollectAccelerometerDatav2.

Current Abilities

The application has the following capabilities:

• Recognition of Different Actions: It can recognize various actions such as running, jogging, standing, cycling, etc.
• Segmentation of Actions in a Routine: The collected data can be segmented to identify different actions within a routine.

Future Features

The project aims to implement the following features in the future:

• Improved Speed and Accuracy of the Model: Enhancements will be made to optimize the speed and accuracy of the action prediction model.
• Mobile Device Compatibility: The goal is to enable the entire process to run seamlessly on a mobile device.

Usage Guide

To add a new action to the application, follow these steps:

1. Record Data: Execute /recordData.py to record data from your mobile device. Connect the device and wait for approximately 30 seconds to collect accelerometer data. The recorded data will be saved in yyyymmddhhmmss.csv.

2. Clean and Refine Data: Run /processData.py to clean and refine the collected data. You can also label the data at this stage.

3. Train the Model: Execute /model.py to train the model using the cleaned and labeled data.

Once the model is trained, you can:

• Run the Application: Execute ./main.py to start the application. Connect your device and enjoy the action prediction capabilities!