How badie16/phytovigil works
Project type and purpose
• PhytoVigil is a full-stack plant-disease-detection solution comprising three main user-facing clients (mobile, web admin) and a backend API with an embedded AI service.
• Key use cases: farmers snap leaf photos on mobile, images are sent to the backend for CNN inference, results and natural-treatment recommendations are returned; administrators manage diseases, plants, users via a Next.js admin UI.
File structure
• Top level:
• backend/ – FastAPI monolith (app/core, crud, routes, services, ml, models, schemas)
• mobile/ – React Native (Expo) application (app screens, components, hooks, contexts, services for local/remote storage)
• (README.md points to a web/ folder for Next.js admin UI, not shown in file tree; assume similar structure under web/)
• backend/app/core: config, security → app bootstrap and JWT setup
• backend/app/crud + models + schemas → ORM layer (SQLAlchemy + Pydantic)
• backend/app/routes → API endpoints (auth, ml, plants, scans, diseases, users, files)
• backend/app/services → domain logic (file uploads, ML processing)
• backend/app/ml → model artifacts (H5 file), preprocessor, loader, prediction service
• mobile/app & mobile/components & mobile/services → UI screens, reusable UI primitives, offline (SQLite + AsyncStorage) and remote services for API calls
README for additional insights
• Confirms technology stack: FastAPI, PostgreSQL, TensorFlow, JWT on backend; React Native + AsyncStorage on mobile; Next.js + Tailwind CSS for admin web; Gemini API for extra info.
• Highlights feature set: secure auth with roles, local/cloud sync, AI-driven disease detection, admin management console.