✨ Overview
AgriBuddy is an AI-powered, agentic support system designed to provide personalized and real-time agricultural guidance to farmers in Bangladesh. It combines advanced language and vision models, including a Retrieval-Augmented Generation (RAG) framework, to deliver Bangla-language recommendations through a smart chatbot.
AgriBuddy supports:
Natural Bangla or Banglish text queries
Image-based rice disease detection using Convolutional Neural Networks (CNN)
Context-aware farming advice tailored to the user's region, crop, and conditions
🚀 Core Features
Multilingual Query Understanding: Comprehends queries in Bangla, English, and Banglish.
Rice Disease Identification: Detects 10 classes of rice conditions from images (9 diseases and 1 normal case).
Expert Agricultural Advisory: Provides guidance via an integrated knowledge base.
Personalized Recommendations: Utilizes user profiles and memory agents for tailored advice.
Integrated Tools: Offers features for weather analysis, soil condition assessment, and cultivation tips.
🎯 Built For
AgriBuddy is specifically built for smallholder and rural farmers in Bangladesh who require accessible, expert-level agricultural support without the burden of traveling long distances.
⚙️ Technical Highlights
Modular Agent-Based Design: Features Smart Query, Memory, and Expert Agents for robust functionality.
CNN-Based Disease Detection: Trained on the specialized Paddy Doctor dataset for high accuracy.
RAG-Powered Answer Generation: Leverages Bangla embeddings for contextually relevant responses.
Deployment: Implemented as a mobile-first Progressive Web App (PWA) with built-in offline support.
📊 Dataset and Code Access
Rice Variant & Disease Dataset (Please replace with actual link)
Project Code Repository (Please replace with actual link)
👥 Developed By
Md. Shaleh Islam Tonmoy
Md. Rezuwan Hassan
Tanmoy Shome
Rawhatur Rabbi
Ruwad Naswan
Affiliations
BRAC University
Bangladesh University of Engineering and Technology
Daffodil International University
📺 Presentation & Report
