An AI-powered Sports Injury Risk Detection system using Computer Vision, Pose Estimation, Biomechanics and Machine Learning.
This project aims to detect potential sports injuries before they occur by analyzing an athlete's body movements from video.
The system extracts body landmarks using Google's MediaPipe Pose Landmarker, calculates biomechanical joint angles, analyzes movement patterns, and predicts injury risk using machine learning.
- 🎥 Video Processing
- 🧍 Human Pose Estimation
- 📍 Landmark Extraction
- 📊 CSV Export of Body Keypoints
- 🦴 Skeleton Visualization
- 📐 Joint Angle Calculation
- 🦵 Knee Angle Tracking
- 🤖 Injury Risk Prediction (Upcoming)
- 📈 Dashboard Visualization (Upcoming)
- Python
- OpenCV
- MediaPipe Tasks API
- NumPy
- Pandas
- Scikit-Learn
- TensorFlow
- Git
- GitHub
Sports-Injury-Risk-Detection
│
├── Milestone 1
│
├── Milestone 2
│ ├── backend
│ │
│ └── pose_estimation
│ ├── video_reader.py
│ ├── pose_detector.py
│ ├── landmark_extractor.py
│ ├── csv_exporter.py
│ ├── visualize_skeleton.py
│ ├── joint_angle_calculator.py
│ ├── calculate_knee_angles.py
│ └── angle_exporter.py
│
├── models
│
└── outputs
Project setup
Completed Pose Estimation Pipeline
✔ Video Reader
✔ Pose Detection
✔ Landmark Extraction
✔ CSV Export
✔ Skeleton Visualization
✔ Joint Angle Calculation
✔ Knee Angle CSV Export
- Hip Angles
- Elbow Angles
- Shoulder Angles
- Biomechanical Features
- Gait Analysis
Pose Detection Accuracy
- Frames Processed: 484
- Detection Rate: 100%
Landmark Extraction
- Total Landmarks: 15,972
Joint Angle Extraction
- Knee Angles exported for all frames.
- Injury Risk Prediction
- Deep Learning Models
- Athlete Dashboard
- Real-Time Webcam Detection
- Injury Alerts
- Performance Analytics
Rachit Patnaik
B.Tech Computer Science (Data Science)
ITER, SOA University
GitHub: https://github.com/Rachit-Patnaik
LinkedIn: https://linkedin.com/in/rachit-patnaik-87933a332
⭐ If you like this project, consider giving it a Star.