How nadaelmasry/cephalometric-landmark-detection works
Main Components
• Frontend Client – The “frontend” directory contains HTML, CSS, and JavaScript files (including index.html, upload.html, result.html, etc.). This is used for user interactions (uploading patient images, viewing reports, help pages, etc.).
• Node.js API – The “node.js Api” folder indicates your backend Node.js application with folders for DB connection, models, controllers, routes, and services (e.g., cloudinary integration). It handles operations such as adding new patients, managing medical records, and interfacing with the MongoDB database.
• Flask API – The “flask api” directory holds several Python files and Jupyter notebooks that are used for deep learning operations (landmark positioning using UNet, Inception, and ResNet models), indicating the segmentation/detection part.
• Database – Though not a folder, the README specifies MongoDB as the database for storing patient data and records.
• Deep Learning Model – The “models notebooks” folder shows notebooks for training and experimenting with different models (binary mask UNet, Inception, ResNet). These represent the ML processes that are deployed via the Flask API.
Relationships and Interactions between Components
• The Frontend interacts with the Node.js API:
• The Node.js API communicates with the Database (MongoDB):
• The Node.js API integrates with the Flask API:
• The Flask API runs the deployed deep learning models:
• Internally, the Flask API leverages TensorFlow, OpenCV, and NumPy to perform image processing and landmark predictions. Indicate the incorporation of these technologies as annotations or labels near the Flask API component.
Architectural Patterns and Design Principles
• Separation of Concerns:
• Microservices or Service-Oriented Architecture:
• MVC-like pattern in Node.js API:
• Within the node.js Api folder, emphasize the controllers, routes, and models which indicate an MVC-inspired structure.
• Integration of ML Workflows: