Overview

The virtual personal fitness trainer developed during HackNYU2018 by Jason Chin, Charlie Lin, Brad Huang, and Calvin Woo is an impressive project that leverages artificial intelligence and computer vision to enhance workout efficiency. Built in a mere 36 hours, this initiative focuses on guiding users through exercises using real-time feedback, making it a revolutionary tool for fitness enthusiasts. The application aims to critique user form and count repetitions with impressive accuracy, utilizing advanced technology such as 2D human pose estimation.

While the project holds great potential, it’s important to note that it hasn't been actively maintained over the past three years. Users should approach the functionality with this context in mind, as updates and refinements are essential for integration with evolving web technologies and exercise methodologies.

Features

2D Human Pose Estimation: Utilizes advanced algorithms to track user joints through a standard webcam, providing real-time critiques of exercise form.
Web-Based Application: Runs effortlessly in the browser, allowing for accessibility across devices equipped with a camera and a WebRTC-supported web browser.
Real-Time Feedback: Processes webcam feed to offer immediate feedback on exercise execution, enhancing the workout experience by correcting bad form.
Exercise Variety: Currently supports analysis for several exercises, including squats, dumbbell shoulder presses, and bicep curls, with specific emphasis on form errors.
Future-Proofing Potential: Envisions advancements such as support for multiple users and refined pose estimation for improved accuracy, inviting future development.
Flask Server Integration: Performance is bolstered by a Flask server that processes the data, allowing for robust functionality that can scale to different hardware setups.
Post-Processing Option: Includes the ability to send recorded workout videos to the server, accommodating users with slower network speeds for a consistent experience.

Jrobchin

Phormatics