This repository contains all the codes used for data cleaning, data processing, model training, and model deployment for the thesis "Machine Learning Based Real-time Movement Detection of Children" (BSc Thesis, 2024, ELTE IK).
The dataset utilized in this project is derived from a study that involved 40 children, with the goal of recognizing various activities through wearable devices. The data, structured in pickle files and accompanying CSV files for labels, provide a comprehensive basis for training our machine learning models.
The dataset includes:
- Pickle Files: Contain raw sensor data from wearable devices, capturing movement specifics of each child involved in the study.
- CSV Files: Accompany the pickle files and contain labels for the movements. Each label corresponds to a specific activity recorded in the pickle files.
The methodology and detailed description of how the data was collected can be found in the following research paper: