BoneVision AI is our hands-on project where we took a pre-trained YOLOv8 model and fine-tuned it ourselves on 8K+ real X-ray images to detect bone fractures. Doctors can upload an X-ray, and within seconds get bounding boxes around fractures, confidence scores, and helpful medical insights.
The system combines cutting-edge machine learning technology with a user-friendly interface, allowing doctors and radiologists to simply upload X-ray images and receive immediate AI-assisted analysis with confidence scores and medical recommendations.
├── data/
│ ├── raw/ # Train daata
│ │ └──images/ # Original X-ray images (~8k+)
│ │ └──labels/ # YOLO annotation files (.txt)
│ └── processed/ # Preprocessed images & augmentations
├── models/
│ ├── pretrained/ # Base YOLOv8 weights (yolov8n.pt)
│ └── trained/ # Fine-tuned fracture models
├── results/
│ ├── metrics/ # mAP, precision, recall plots/JSON
│ └── predictions/ # Sample detection outputs
├── static/ # .gitkeep
├── uploads/ # User-uploaded images (Flask)
├── .gitignore # Python/ML ignores
├── version.json # App & model versions
├── README.md # This file
├── app.py # Flask web server
├── bone_fracture_detector.py # Core YOLO detection
├── bone_fracture_detection2.py #new detection and train code
├── test.py # Testing code for dataset and runner code and dir check
├── test2.py # new test script
├── index.html # Basic html code
├── styles.css # css code
├── script.js # js code
├── streamlit_site.py # py script for dashbord with backend for this application
├── fine_tunning_algo.cpp # cpp code for the fine tunning the model
└── requirements.txt # Dependencies
Backend: Python • Flask • YOLOv8 (fine-tuned) • PyTorch • OpenCV • Ultralytics
Frontend: HTML5 • CSS3 • JavaScript • Font Awesome • streamlit-py
Deployment: ...
Clone the repository
bash git clone https://github.com/nabakrishna/bone-fracture-detection.git cd Bone-Fracture-Detection Install dependencies
bash pip install -r requirements.txt Run the application
bash python app.py Open your browser
text http://localhost:5000 🎯 How It Works Upload an X-ray image through the web interface
AI model processes the image and detects potential fractures
Results display original image alongside annotated version with bounding boxes
Get detailed analysis with confidence scores and medical recommendations
📋 Features
✅ Real-time fracture detection
✅ Visual results with bounding boxes
✅ Confidence scoring
✅ Medical recommendations
✅ Responsive web design
✅ Privacy-focused (local processing)
This tool is designed to assist healthcare professionals and should not replace professional medical diagnosis or treatment decisions.
MIT License - feel free to use and modify for your projects.