New2Fit is a personalized AI fitness application for beginners featuring an AI camera with real-time pose analysis and feedback.
New2Fit Link: https://new2fit.vercel.app
New2Fit uses a TensorFlow pose model (MoveNet) to analyze exercise form in real time and provides per-rep feedback on range of motion, speed, and joint angles. The application features a modern full-stack architecture with a React (Vite) frontend and a FastAPI backend. Real-time video processing is handled via WebSockets, and user data is managed securely using SQLAlchemy.
- Real-time AI Camera Body Tracking: Analyzes video frames using MoveNet via WebSockets.
- Form Analysis & Feedback: Provides per-rep feedback, angle measurement, and speed monitoring.
- Modern Web Interface: Built with React and TypeScript for a highly responsive user experience.
- Workout Logging: Track your exercises, sets, reps, and weights over time.
- Personalized Dashboard & Questionnaire: Customize your fitness journey based on your personal preferences and goals.
- Real-time Chat & User Communication: Instant user-to-user messaging powered by WebSockets, complete with live unread message badges and active conversation state syncing.
- Gym Locator: Gym locations and websites retrieved from a web scraping algorithm.
- Google OAuth: Secure user authentication.
- Frontend: React, TypeScript, Vite, React Router
- Backend: FastAPI, Python, WebSockets, SQLAlchemy, Pydantic
- AI/ML: TensorFlow (MoveNet)