A Smart Yoga App that uses AI to provide real-time feedback and guidance for anyone looking to learn yoga or perfect their form.
🏆 Submitted to BrickHack 11
YogiSync leverages cutting-edge AI and computer vision technologies to create an all-purpose yoga guidance solution. Whether you're a beginner looking to learn proper form or an experienced practitioner seeking to refine your technique, YogiSync provides personalized, real-time feedback anytime, anywhere.
The inspiration for YogiSync came from our own challenges with maintaining proper form during yoga practice. We noticed that many practitioners, especially beginners, struggle to self-correct without immediate feedback from a teacher. Combining our passion for technology with a love for yoga, we envisioned a tool that could offer personalized guidance—making yoga more accessible and effective.
- 🎥 Real-Time Pose Detection: Uses MediaPipe to detect and track body landmarks from your webcam feed
- 🤖 AI Pose Classification: Custom-trained ML model (Random Forest) classifies various yoga poses with high accuracy
- 📊 Instant Feedback: Overlays skeleton visualizations on live video and provides corrective instructions for proper alignment
- 💬 AI Yoga Coach Chatbot: Powered by Google's Gemini LLM, recommends ideal yoga poses based on your needs and goals
- 📝 Personalized Plans: Generates customized yoga and dietary routines tailored to individual users
- 🎵 Playlist Creation: Creates yoga pose sequences optimized for your fitness level and objectives
- Python - Core programming language
- Flask - Web framework for API endpoints and request handling
- MediaPipe - Real-time pose landmark detection
- scikit-learn - Machine learning model training (Random Forest)
- Pickle - Model serialization
- HTML/CSS - Structure and styling
- JavaScript - Interactive functionality
- Bootstrap - Responsive UI framework
- Google Gemini LLM - Conversational AI yoga coach
- Custom ML Model - Pose classification trained on diverse dataset
- MediaPipe Pose - 33-point skeletal tracking
- Python 3.8 or higher
- pip package manager
- Webcam (for pose detection)
-
Clone the repository
git clone https://github.com/yourusername/yogisync.git cd yogisync -
Create a virtual environment
python -m venv venv source venv/bin/activate # On Windows: venv\Scripts\activate
-
Install dependencies
pip install -r requirements.txt
-
Set up environment variables
cp .env.example .env # Add your Gemini API key to .env -
Run the application
python app.py
-
Open your browser and navigate to
http://localhost:5000
- Start a Session: Allow camera access when prompted
- Select a Pose: Choose from the available yoga poses or ask the chatbot for recommendations
- Follow Instructions: Position yourself in front of the camera and attempt the pose
- Receive Feedback: Watch the skeleton overlay and read real-time corrective guidance
- Track Progress: Monitor your form improvements over time
- Consult AI Coach: Chat with the AI coach for personalized routines and dietary advice
We collected a custom dataset of yoga poses featuring diversity in body types, lighting conditions, and camera angles to ensure robust model performance.
Webcam Feed → MediaPipe Pose → 33 Landmark Coordinates → Preprocessing
Landmarks → Feature Extraction → Random Forest Model → Pose Classification
- Calculate joint angles for key body positions
- Compare against ideal pose parameters
- Generate specific corrective instructions
- Overlay visual feedback on video stream
- Process user queries through Gemini LLM
- Generate contextual recommendations
- Create personalized workout plans
- Provide nutritional guidance
| Team Member | Role | Contribution |
|---|---|---|
| Tanvi Chandan | UI/UX Design & Frontend | Designed intuitive interfaces with strong focus on usability and led frontend integration |
| Anusha Seshadri | Full Stack Development | Developed LLM-based yoga coach and chatbot, contributed to frontend and backend integration |
| Shubh Sehgal | Backend Lead | Spearheaded backend development, integrated MediaPipe, trained ML models (Random Forest achieved lowest test loss), and developed APIs |
| Iyashi Pal | Backend Development | Integrated custom ML model for pose classification, implemented corrective feedback logic, and managed MediaPipe pipeline |
- Google MediaPipe team for the excellent pose detection library
- BrickHack 11 organizers and mentors
- The yoga community for inspiration and feedback
- Google Gemini for powering our AI coach
Project Link: https://devpost.com/software/yogisync
Made with ❤️ at BrickHack 11