Skip to content

Home

Jump to bottom
George-Kariuki edited this page Nov 4, 2025 · 1 revision

🌾 AgriScan Wiki

Welcome to the AgriScan Project Wiki — the knowledge base for contributors, developers, and partners helping us build the future of AI-powered crop health diagnostics.

🧭 Overview

AgriScan is an AI-powered mobile crop-health scanner for smallholder farmers.
It detects plant diseases using a simple smartphone camera and provides instant treatment guidance — even in offline or low-connectivity environments.

Mission: Empower farmers with affordable, instant crop diagnostics through mobile AI.

🏗️ System Architecture

AgriScan follows a modular, scalable architecture that combines a lightweight mobile client, FastAPI backend, and TensorFlow AI model.

Component Tech Description
Mobile App Flutter Captures leaf images, runs on-device inference, syncs with backend
Backend FastAPI + Python Handles API requests, predictions, and database sync
Database Supabase / PostgreSQL Stores users, scan logs, and image metadata
AI Model TensorFlow / PyTorch Classifies crop diseases and generates confidence scores
CI/CD GitHub Actions + Docker Automates builds, testing, and deployment

📘 See: Architecture Diagram

🧪 Current Status

Version Stage Focus
v0.1 MVP Crop scan + disease detection + offline logbook
v0.2 (Planned) Pilot Release Treatment advice, SMS alerts, and admin tools

✅ Completed

  • TensorFlow Lite model integration
  • FastAPI /predict endpoint
  • Flutter mobile app with camera + Hive offline storage
  • Automated API + Flutter tests
  • CI/CD with GitHub Actions

🚧 In Progress

  • Treatment recommendation logic
  • SMS integration
  • Admin dashboard

🧠 Core Features

  • 📸 AI Disease Detection – Snap a photo, get instant diagnosis
  • 💊 Treatment Suggestions – Context-aware recommendations (coming soon)
  • 📓 Farm Logbook – Offline image + scan history
  • 📶 Offline Sync – Works without internet, syncs when online
  • 🧑‍🌾 Multi-Crop Support – Starting with cassava; expanding to maize, tomatoes, and beans

📘 See: Feature Roadmap

⚙️ Setup Guide

Follow these quick-start guides to run AgriScan locally:

🧩 AI Model Details

The disease classification model is based on TensorFlow Lite, trained on cassava datasets from open agricultural sources and fine-tuned for mobile efficiency.

📘 See: Model Training & Conversion

📈 Public Roadmap

Quarter Milestone Goal
Q1 Research + UI + Data Collection Build datasets and base UI
Q2 AI Inference + MVP App Achieve end-to-end inference
Q3 Pilot with Farmers Field testing and feedback
Q4 Monetization Partner with co-ops and agritech firms

🧑‍💻 Contributing

We welcome contributors! Whether you’re into AI, Flutter, or backend systems — there’s a space for you.

How to contribute:

  1. Fork the repo and clone locally
  2. Create a feature branch
  3. Submit a PR with clear commit messages

📘 See: Contributing Guidelines

📄 License

This project is licensed under the MIT License.
See the main repo’s LICENSE file for details.

🌍 Community & Links

🪴 Next Steps

After reading this page, explore:

Clone this wiki locally