AI analysis of KU-HAR human activity recognition data

[ with: Harshit Jaglan, Thuan Lam, Ynha Nguyen, Thy Tran ] This project uses the KU-HAR human activity recognition dataset to train models that classify movements from wearable sensors. It explores preprocessing, feature engineering, and supervised learning techniques. Project work: - Sensor stream preprocessing and noise reduction - Feature extraction across temporal windows - Training classifiers for activity labels like walking, sitting, and jogging The aim is to build a practical activity recognition pipeline that can be applied to health, fitness, and ambient sensing solutions.