The Virtual Physiotherapy Assistant is a web-based platform that utilizes AI-powered pose estimation and chatbot support to assist patients in rehabilitation exercises. The system provides real-time feedback, and mental health support to enhance adherence and recovery
- Pose Estimation & Real-Time Feedback
- Utilizes Yolov11 for accurate body movement tracking
- Provides real time visual overlays and alerts user for incorrect form
- AI-Powered Physiotherapy Chatbot
- Offers motivational support and encouragement
- Context-aware respones tailored to user engagement
- Web-based Accessibility
- TypeScript frontend with FastAPI backend
- Supports real-time webcam tracking
| Component | Technology |
|---|---|
| Frontend | Next.js / Tailwind CSS |
| Backend | FastAPI |
| Pose Estimation | Yolov11 |
| Chatbot AI | OpenAI GPT |
| Database | PostgreSQL |
flowchart LR;
step1["Patient uploads a video or enables webcam, and selects a exercise"]
step2["Yolov11 detects keypoints in real-time"]
step3["System evaluates pose accuracy and provides feedback"]
step4["User may access the chatbot to request simpler exercises or request for more information regarding the current exercises"]
step1 --> step2;
step2 --> step3;
step3 --> step4;
| Challenges | Solution |
|---|---|
| Accurate pose tracking | Adjusted the skeleton model for more accurate tracking |
| Documenting Feedback | Utilized OpenAI for concise and clear feedback formatting |
| Early Rep Repetition Misinterpretation | Creating a null zone where errors would not be counted |
- Clone the repository
git clone https://github.com/andrew-juang/Stride.git cd Stride - Install the dependencies
cd web_app npm install cd .. pip install -r requirements.txt - Start the frontend development server
npm run dev - Navigate to the backend and install a Postgresql database
CREATE DATABASE stride_db; - Run the backend
uvicorn backend.main:app --reload - Navigate to http://localhost:3000/
For more information, please contact Andrew Juang at [email protected]