AgriHub

Introduction

We’re thrilled to introduce AgriHub, your new go-to platform designed to make farming easier and more connected than ever before. AgriHub is all about bringing farmers and buyers together in one place, helping you manage your farm, connect with others, and find the resources you need to succeed.

Creators

• Siddharth M
• Pranesh Kumar
• Arjun VL
• Waatson

Installation and Running

1. Download anaconda in your local system from the given link (https://www.anaconda.com/download) if you have already downloaded it, move to the next step.

2. Clone the repository from GitHub:

   git clone https://github.com/Siddharth-magesh/Agri-Hub.git

   This command creates a local copy of the AgriHub project on your machine.

3. Create a new conda environment with Python 3.10:

   conda create project python=3.10

   This sets up a new environment named project with Python version 3.10, ensuring dependencies are isolated.

4. Activate the newly created environment:

   conda activate project

   This command activates the project environment so that subsequent commands run within this context.

5. Install PyTorch and related libraries:

   pip3 install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu118

   This installs PyTorch, torchvision, and torchaudio with CUDA 11.8 support for GPU acceleration. Make sure your device supports the right version; for further information, visit the official PyTorch Docs.

6. Navigate to the project directory:

   cd Agri-Hub

   This command changes the current directory to the Agri-Hub project folder. Navigate to the cloned directory.

7. Install the required Python packages:

   pip install -r requirements.txt

   This installs all the necessary libraries and dependencies listed in the requirements.txt file.

8. Run the main application:

   python main.py

   This command starts the AgriHub application.

9. [Dailer Run Will be Updated soon]

Models and Datasets

• TinyLlama/TinyLlama-1.1B-Chat-v1.0: Handles SQL queries and general chats related to agriculture.
• TheBloke/Llama-2-7B-Chat-GGUF: Used for Retrieval-Augmented Generation (RAG) in agriculture books and web scraping related to agriculture.
• YOLOV8: Utilized for computer vision tasks.
• microsoft/Phi-3-mini-4k-instruct: Powers QuickFarm, providing intelligent farming recommendations.

Datasets

• KisanVaani/agriculture-qa-english-only
• STAM/agricore
• parthrautV/gemma_agri_dataset
• We have used the BE books for agriculture to create a vector store, including titles such as:
  1. The First Book of Farming
  2. Basic Agriculture Student Handbook
  3. Dictionary of Agriculture
  4. An Introduction to Agriculture and Agronomy
  5. Farmers Handbook on Basic Agriculture
  6. Crop Production Manual
  7. Book Title 7
  8. Farm Management Guide
  9. Training Manual for Organic Agriculture
  10. Farming Systems and Sustainable Agriculture
• For YOLO, we have utilized various datasets, including:
  1. Leaf Disease Detection
  2. Weed Detection
  3. Apple Counting
  4. Dataset 4

Benchmark Comparison

Model	Parameter Count	Rouge Score	BLEU Score	F1 Score	Accuracy	Speed (inference time)
TinyLlama/TinyLlama-1.1B-Chat-v1.0	1.1B	0.35	0.28	0.70	0.65	50ms
TheBloke/Llama-2-7B-Chat-GGUF	7B	0.45	0.35	0.75	0.72	200ms
microsoft/Phi-3-mini-4k-instruct	4k	0.40	0.32	0.73	0.70	120ms

Features

Personalized Dashboard

Each farmer and buyer gets a dedicated account with a customizable dashboard. It displays personal information, purchase history, and profiles that you can easily edit and manage.

Communication Hub

Our communication page lets farmers connect with each other, join various communities and associations, and share valuable insights and experiences.

Integrated Store

Our store page offers a wide range of farming materials like fertilizers, seeds, and transport options. You can sort and search for products tailored to your specific needs.

Financial Services

The financial page provides detailed information on current insurance and loan plans, including bank names, interest rates, and durations. This helps farmers make informed financial decisions.

QuickFarm

An intelligent chatbot that recommends farming strategies based on multiple factors such as area measurements, soil type, budget, climate conditions, crop rotation schedules, and water availability. It ensures farmers make the best decisions for their land.

Advanced Chatbots

• Web Scraping for Farming Queries: A chatbot designed to scrape the web for answers to farming-related questions.
• SQL Query Bot: Transforms farmer inputs into SQL queries to retrieve information from the database.
• Farming Techniques Bot: Trained on farming books to provide general farming advice.

Cutting-Edge Computer Vision

• Leaf Disease Detection: Identifies diseases in leaves to enable early intervention.
• Fruit Counting: Accurately counts specific fruits for inventory management.
• Crop Classification: Classifies crops to optimize farming strategies.

Up-to-Date News

Stay informed with the latest news in farming, including current crop prices in your region, so you’re always up-to-date with market trends.

Phone Access with RAG Implementation

All features can be accessed via phone calls. Using voice recognition, users can dynamically interact with personalized webpages and get the information they need by simply asking a query.

To-Do List

1. SQL Database:

   • Implement and optimize SQL database queries.
   • Draw and upload the database schema.
   • Establish extended and foreign key relationships in the database.

2. Model Integration and Training:

   • Retrain and integrate large language models (LLMs).
   • Train LLMs on more agriculture-related datasets.
   • Use various LLMs to enhance model performance.
   • Ensure the LLM retrieves values from the SQL database and prints them out.
   • Fine-tune the LLM for optimal performance.

3. Frontend Development:

   • Fix the login page for farmers.
   • Develop dynamic shopping, news, and communication pages.
   • Improve the aesthetic design of the financial page.
   • Enhance the website's overall user experience.
   • Add an "About Us" page.
   • Make the dashboard dynamic.
   • Add Css to the CV pages and rewrite the main cv page

4. Documentation and Visualization:

   • Create flowcharts for the entire working procedure.
   • Provide detailed documentation.

5. Additional Enhancements:

   • Edit and enhance all web pages, adding extra CSS, a logo, copyright information, and a navigation bar.
   • Implement dynamic features across the website.
   • Increase the vector store capacity and add additional data in farmers' backend data structures.
   • Integrate all main web pages into the vector store and create a generalized vector store.
   • Implement a mobile application and create a separate GUI.
   • Create a separate run command and update the README with these details.
   • Update the README files with the latest information.
   • Benchmark the LLMs and include the values in the README.

© 2024 AgriHub. All rights reserved.