-
-
MentorX - A Free Personal Trainer By Your Side
-
Example User Interface
Inspiration
While maintaining a fitness lifestyle is highly popular, a common—and free—solution is simply imitating online influencer videos. However, I realized through my own fitness journey that without a coach, pushing toward demanding physical goals can easily lead to poor form and injury. The help from pre-recorded videos is strictly one-way and limited.
This is the motivation behind Mentor-X. The value of Mentor-X lies in its accessible, AI-driven advisory service, leveraging computer vision to detect posture through any standard camera device. More importantly, in the context of Generative AI, Mentor-X takes that raw posture data and generates personalized, conversational feedback. It allows users to review their form, understand exactly where they need to improve, and receive dynamically generated workout adjustments—delivering the precise guidance of a personal fitness trainer, completely free of cost.
What it does
Mentor-X is an AI-assisted web application that acts as a virtual, on-demand fitness coach. Designed to provide an immediate solution for users without access to an in-person trainer, the platform requires nothing more than a Wi-Fi connection and a standard device camera. Under the hood, Mentor-X creates a seamless user experience by combining two powerful technologies. First, it leverages Google's open-source MediaPipe framework to perform real-time gesture detection, analyzing the user's movements to determine if a repetition is performed with a safe, successful form. Then, Mentor-X feeds this movement data into Google Gemini to instantly generate and return highly personalized, actionable fitness advice. This ensures users are not just mindlessly completing reps, but actively improving their posture and technique from wherever they choose to work out. MentorX also supports bilingual language options in English and Traditional Chinese.
How we built it
Languages:
HTML - for UI Python - for backend logic
Frameworks and Libraries:
Framework 1: Vue - for building MentorX’s UI Library 1: Pinia - for state management Framework 2: MediaPipe - for gesture detection Library 2: Numpy - for angle calculation
Platforms:
Gemini: to provide fitness advice in natural language
Challenges we ran into
As we delegated our backend and frontend modules to different teammates, it took us a while to coordinate, but it turned out well!
Accomplishments that we're proud of
Easy communication.
What we learned
As a team, we learned how to communicate and delegate tasks effectively under time pressure.
What's next for MentorX
Expanded Athletic Tracking:
During discussions, we see MentorX's potential to adapt our gesture-detection models for specialized activities. This includes high-precision sports like golf, as well as endurance training where sustained, safe posture is essential.
