Pose buddy : workout platform with PoseNet

• Main Author: Ong, Jia Ying
• Other Authors: Owen Noel Newton Fernando
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2021
• Subjects: Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision
• Online Access: https://hdl.handle.net/10356/148205
• Institution: Nanyang Technological University

Due to the Coronavirus outbreak, exercising outdoors has been made near impossible. Some have started working out at home in hopes of keeping fit. However, many lose interest after a while as it becomes mundane, or simply because they are used to working out with friends. Furthermore, being confined in their own homes has resulted in numerous citizens transiting into a sedentary lifestyle. Prolonged inactivity has been shown to cause obesity which correlates with many other health diseases. Hence, this project introduced an online workout platform that promotes exercise by offering interactivity using computer vision and collaboration through connecting with others. Pose Buddy is a web application that primarily uses an in-house Pose-System, where it would provide real-time feedback during each workout session. The Pose-API has a training dataset of over 500 entries for different types of exercise. These data were collected from numerous sources, namely Yoga-82, Unsplash, and self-taken images of each pose. Google's Teachable Machine was used to train the machine learning model for the poses. With the aforementioned, real-time inputs from the camera feed can be captured and processed asynchronously by the system, allowing users to know in real-time if they are carrying out the workout correctly. Additionally, this application can support most browsers and operating systems.