This project was developed as part of the PIDEV – 2nd Year Engineering Program at Esprit School of Engineering – Tunisia (Academic Year 2025–2026). Main Technologies: FastAPI, React, PostgreSQL, Docker, Unity (Digital Twin)..
🌱 SanIA — Smart Agriculture Intelligence
Best Project Award — Data Science Engineering, ESPRIT 2025–2026 🏆
SanIA is an end-to-end precision farming platform that combines AI, IoT, and 3D digital twin technology to help farmers monitor crops, detect diseases, automate irrigation, and get intelligent agronomic advice — all from a single system.
🎥 Demo Video
youtu.be
🚀 Key Features
ModuleDescription🌿 Crop Disease DetectionCNN-based mobile app that identifies plant diseases from leaf photos in real time🏗️ 3D Digital TwinInteractive Unity-based 3D model of the farm with live IoT sensor data🤖 RAG ChatbotLLM-powered agricultural assistant using Retrieval-Augmented Generation💧 Smart IrrigationIoT-automated irrigation system triggered by soil and weather sensor data🛰️ Satellite + Sensor MonitoringField health monitoring combining satellite imagery and ground-level IoT sensors📱 Mobile AppReact Native app for disease detection, chatbot access, and IoT dashboard
🛠️ Tech Stack
AI / ML
Computer Vision (CNN) — crop disease classification
RAG pipeline — LLM + vector database for agronomic Q&A
FastAPI — model serving backend
Frontend & Mobile
React Native — cross-platform mobile app
Unity — 3D digital twin visualization
Backend & IoT
Node.js / Express
MQTT / IoT sensor integration
Satellite data processing
📁 Project Structure
sanIA/
├── mobile-app/ # React Native app
├── digital-twin/ # Unity 3D farm model
├── disease-detection/ # CNN model + FastAPI
├── rag-chatbot/ # LLM chatbot backend
├── iot-irrigation/ # IoT automation scripts
└── backend/ # Main API server
👥 Team
4th Year Data Science Engineering — ESPRIT, Group 4DS10
Academic Year 2025–2026
🏆 Recognition
🥇 Selected as one of the best projects of the Data Science Engineering department — ESPRIT Bal des Projets 2026
Built with ❤️ for Tunisian agriculture