🐄 JAGRUTHI - Livestock Breed Identification App

Empowering Farmers with AI-driven Breed Detection

🚀 Overview

JAGRUTHI is an AI-powered mobile application built using React Native and PyTorch to accurately identify livestock breeds from images. The app is designed to assist farmers, veterinarians, and agricultural officers in recognizing various cattle and goat breeds, promoting better care and breeding strategies.

• 📱 Cross-platform mobile app
• 🧠 EfficientNet-B2 based breed classification
• ☁️ Dockerized backend deployed on Google Cloud
• 🤝 Government-partnered initiative launched by the Agriculture Minister in Bidar

---

🎯 Features

• 📷 Upload or Capture Livestock Images
• 🧬 Accurate Breed Prediction (96.3% Accuracy)
• 🌍 Scalable & Reliable Deployment (99.9% Uptime)
• 📊 Breed Details & Characteristics Displayed
• 🖼️ Intuitive UI with Localization (Kannada Support)

---

🛠️ Tech Stack

Frontend	Backend	Model Training	Infrastructure
React Native	Node.js + Express	PyTorch + EfficientNet-B2	Google Cloud Run + Docker

---

🧠 Model Highlights

• 🏗️ Architecture: EfficientNet-B2 (pretrained)
• 🧪 Accuracy: 96.3% on real-world validation set
• 🗃️ Classes: Bidari, Deoni, Girr, Kenguri, Murra, Sirohi
• 📊 Metrics:
  • Precision: X.XXX
  • Recall: X.XXX
  • F1 Score: X.XXX

📁 Model Training Code (Colab) | 📦 Saved Model

---

📸 App Screenshots

Home	Upload	Prediction

---

🏛️ Government Partnership

The app was proudly launched in January 2025 in Bidar, Karnataka in partnership with the Department of Agriculture.

📬 Contact

If you'd like to collaborate, feel free to connect!

Sunil Gopal C V