-
-
Our Team!
Inspiration
💡 Our inspiration came from recognizing a common challenge that many people face when starting their fitness journey — the fear of not knowing what to do or how to do it properly in the gym 🏋️♀️. This uncertainty often leads to discouragement and even giving up. We wanted to change that. Our goal was to create an AI-powered gym companion that makes fitness more approachable, empowering , and personalized.
By integrating computer vision, our app can analyze movements and provide real-time feedback on form, helping users build confidence and avoid injury . We believe that no one should feel lost or alone in their journey toward health , and with this tool, we hope to make the gym a more inclusive , fun, and supportive space for everyone — from beginners to seasoned athletes.
What it does
💪 What it does
Our project is an AI-powered fitness assistant designed to make working out more accessible, personalized, and engaging. At its core, we've built a friendly, interactive chatbot powered by Google’s Gemini AI that guides users through tailored workout plans based on their individual goals, fitness levels, and preferences. Whether someone is looking to build muscle, improve endurance, or simply get started, our chatbot suggests routines, explains techniques, and offers motivation — all in a conversational, approachable way.
To take things further, we integrated a text-to-speech feature that brings the chatbot to life. Users can choose from fun voice tones like “gym bro 💪”, “gym girl💅”, or “neutral🎩” via a dropdown menu on our website — making the experience more entertaining and accessible for those who prefer listening over reading.
But what truly sets our project apart is the incorporation of computer vision to analyze user form in real time. Using pose detection technology, our app evaluates two foundational exercises — the squat 🏋️ and the push-up 🙇. As users perform these movements in front of their camera, our system overlays color-coded landmarks and provides a score based on technique and alignment. This immediate feedback helps users correct their form, avoid injury, and train smarter.
Together, these features form a powerful, AI-driven gym companion that supports users both mentally and physically on their fitness journey. By removing barriers and offering intelligent, responsive guidance, our app makes the gym feel less intimidating — and a whole lot more empowering 🚀
How we built it
🛠️ How we built it
We divided our team into two core groups: back-end developers and front-end developers, working in parallel to bring our web app — Gymini — to life.
On the back end, our developers focused on implementing the core functionality using Gemini AI. Our project serves as a custom wrapper around Gemini, allowing us to build intelligent responses tailored to user goals and conversations. We also integrated MediaPipe to power our computer vision module, which analyzes user form in real time during squats and push-ups. For the text-to-speech feature, we utilized the pyttsx3 Python library to provide voice responses in selectable tones like gymbro, neutral, and gymgirl — making the experience more interactive and motivational.
On the front end, our developers crafted a seamless UI/UX using React ⚛️ and Tailwind CSS. We designed a clean, responsive layout and implemented smooth routing, page transitions, and a navigation bar. The landing page includes an engaging “About Us” section, while the workout page features an integrated camera view to track exercises and display real-time feedback. We also built a color mode switcher, allowing users to choose between different AI personalities for a more personalized coaching experience.
By combining thoughtful design, real-time feedback, and intelligent conversation, we were able to create a full-stack AI-powered fitness assistant that’s both functional and fun.
Challenges we ran into
As first-time hackers, three of us were completely new to the world of development and had never built a full project before 🐣. We faced an initial learning curve with essential tools like GitHub, setting up an integrated virtual environment 🖥️, and understanding the fundamentals of back-end and front-end integration. One of our biggest challenges was working with Python imports and managing the structure of our files while trying to merge separate parts of the code into a single, functioning application ⚙️.
We also encountered difficulties while trying to implement real-time text-to-speech (TTS) feedback 🔊, as integrating Python-based functionality with a web front end proved far more complex than anticipated — especially since none of us had ever built or consumed an API before 🌐. Despite these obstacles, we leaned heavily on mentorship, online resources 📚, and a lot of trial and error to stitch parts of our project together. While some features didn’t make it into the final build due to time constraints and our limited experience, the process taught us a great deal about collaboration, debugging, and adapting quickly under pressure 💪.
Accomplishments that we're proud of
We're incredibly proud to have successfully implemented real-time computer vision in under 12 hours! 🤖⏱️ Even more exciting, three of our team members were first-time hackers 💻✨ — and the amount we learned in such a short time was truly inspiring. From diving into AI models to building full-stack functionality, this project pushed us beyond our comfort zones and proved what we could accomplish together. 🚀💪
What we learned
We learned how to build a multi-page ReactJS app using react-router-dom, manage state with hooks, and style the interface with Tailwind CSS. On the backend, we learned how to set up a Python server, create REST API routes, and handle requests from the frontend. We also gained experience integrating Gemini for generating dynamic chatbot responses 🤖, and processing computer vision data to deliver real-time feedback to users 🎥. Finally, we learned how to merge the frontend and backend seamlessly so that the computer vision and Gemini components communicate effectively with the React app 🔗.
What's next for Gymini
We plan to develop Gymini into a full app with a polished, gym-themed design and mobile compatibility. We also want to expand the voice coaching feature by adding more unique personalities and attitudes for each mode, inspired by characters like Gordon Ramsay, Goofy, and others 🎭. These voices will bring humor, motivation, and variety to the workout experience 💪. Additionally, we hope to improve the chatbot’s fitness advice, refine the computer vision feedback, and create a library of customizable workouts so users can train with Gymini anytime, anywhere 🌟.
Log in or sign up for Devpost to join the conversation.