Workout detection with embedded AI

• Main Author: Ahmad Azfar Bin Abdul Hamid
• Other Authors: Mohamed M. Sabry Aly
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2024
• Subjects: Computer and Information Science
• Online Access: https://hdl.handle.net/10356/181230
• Institution: Nanyang Technological University

The integration of Artificial Intelligence (AI) in wearable technology has transformed fitness tracking, enabling automatic monitoring of activities such as running and cycling. However, traditional fitness trackers struggle with detecting anaerobic exercises like weightlifting, which involve complex and non-repetitive motions. This project addresses this limitation by developing an AI-powered workout detection device capable of recognizing weightlifting exercises, specifically bicep curls, lateral raises, and tricep extensions. Inertial measurement data, including acceleration and angular velocity, were captured using the MPU6050 sensor, which combines an accelerometer and gyroscope. The MAX78000FTHR microcontroller, equipped with a CNN accelerator, was used to classify these movements, enabling efficient workout detection despite hardware limitations. This embedded AI solution enhances fitness tracking, providing more comprehensive monitoring for individuals engaged in strength training.