Govi Isuru 🌾

#Overview

Govi Isuru (ගොවි ඉසුරු — “Farmer’s Fortune”) is a full-stack digital farming platform built for Sri Lankan agriculture. It combines deep-learning crop disease diagnosis, yield forecasting across 25 districts, a peer-to-peer marketplace, weather advisory, and AI-assisted news — with full English & Sinhala support for farmers, buyers, and government officers.

#Highlights

• 🥈 1st Runner-Up & Best AI Team — Devthon 3.0 (University Category)
• 🤖 3 ML Models — Rice (8 classes), Tea (5 classes), Chili (4 classes)
• 📊 10 Years of paddy yield data (2015–2024) across all 25 districts
• 🌐 Full Bilingual — English & Sinhala throughout the product
• 👥 3 User Roles — Farmers, Buyers, Government Officers

#Features

AI Crop Doctor + Grad-CAM

• 🔬 Multi-crop disease detection — Upload leaf photos for Rice, Tea, or Chili diagnosis
• 🗺️ Grad-CAM explainability — Heatmaps show where the model focused
• 💊 Treatment guidance — Bilingual step-by-step recommendations with confidence scores
• 🧠 Transfer learning — MobileNetV2 backbone for efficient on-device-ready inference

Yield Prediction & Analytics

• 📈 District-level forecasting — Predict paddy yield by district, season (Maha/Yala), and year
• 💰 Profit calculator — Revenue, ROI, and break-even yield estimates
• ⚠️ Early warning system — Risk scores with bilingual recommendations
• 🏅 District rankings — Compare all 25 districts by yield, stability, and trend

Marketplace, Alerts & Intelligence

• 🛒 AgroLink marketplace — P2P trading with reputation, ratings, and WhatsApp contact
• 🚨 Community disease alerts — Location-based outbreak warnings at GN Division level
• 📰 Agri news feed — AI summaries, Sinhala translation, and text-to-speech
• 🌤️ Weather advisory — 5-day forecasts with farming-specific recommendations
• 💬 AI crop chatbot — Voice input, conversation memory, and in-chat image diagnosis
• 🌱 Crop suitability advisor — ML recommendations from soil, climate, and irrigation data

Government Officer Tools

• ✅ Report verification & priority management
• 📍 Field visit scheduling with photos and findings
• 📝 Internal notes, audit logs, and performance dashboards

#Tech Stack

#Architecture

govi-isuru/
├── client/        # React + Tailwind frontend
├── server/        # Express + MongoDB API
├── ai-service/    # FastAPI + TensorFlow models
└── docker-compose # Multi-service orchestration

#AI Models

• Rice — 8 classes (Bacterial Leaf Blight, Brown Spot, Leaf Blast, Sheath Blight, and more)
• Tea — 5 classes (Blister Blight, Brown Blight, Gray Blight, Red Rust, Healthy)
• Chili — 4 classes (Leaf Spot, Thrips Damage, Yellow Virus, Healthy)
• Yield Predictor — Trained on 10 years of Sri Lankan paddy statistics across Wet, Dry & Intermediate zones

#Team

• H.M. Kalana Lakshan
• G.H. Lasana Pahanga
• A.M.R. Nawanjana Aththanayake