IoT-Embedded Solution for Agriculture with Machine Learning Technology

Abstract

The project focuses on developing an Artificial Intelligence (AI)-driven Internet of Things (IoT) system for sustainable crop management. It utilizes a machine learning-based crop recommendation engine. The major objective of this system is to optimize crop selection, thereby enhancing yield and operational efficiency. The methodology involves Machine Learning (ML), Deep Learning (DL), and Convolutional Neural Network (CNN) focusing on recommendation and classification techniques.

Key Tools:

• Firebase: For data storage
• Kaggle Datasets: Data source
• Jupyter Notebook: For coding and data analysis
• Pandas: For data preprocessing
• TensorFlow: For model training
• Matplotlib: For visualization

The outcome is a robust recommendation system that identifies the best crop for specific soil conditions and evaluates algorithm performance using metrics and confusion matrices. This project targets farmers, assisting them in the crop selection process by recommending crops based on soil and weather patterns. By automating crop selection, the project saves farmers valuable time while also optimizing yield, ultimately contributing to increased income.

Future Enhancements:

• Real-time monitoring system for soil moisture and sunlight exposure.
• Data analytics for improved crop management.
• User-friendly decision-making interface for AgriSense.
• Plant disease detection AI model.

Visual Representation of Hardware

Here’s a brief overview of the hardware part used in the project, including the IoT setup for soil moisture and sunlight exposure monitoring.

https://github.com/DhruvalAnandkar/AgriScience/blob/main/AgriScience%20Demo%20Video%20(1)%20(1).mp4 => Click Raw

How It Works

1. Sensors: Capacitive soil moisture sensors and light sensors monitor soil conditions and sunlight exposure.
2. Data Collection: Data is sent to ThingSpeak, where it is stored and analyzed.
3. Machine Learning: A recommendation engine uses the collected data to predict optimal crops for specific conditions using TensorFlow and Jupyter Notebooks.
4. User Interface: The system presents recommendations in a simple, easy-to-use interface for farmers to make informed decisions.

Getting Started

Prerequisites

• Python 3.x
• TensorFlow
• Pandas
• Matplotlib
• ThingSpeak API key (for data storage)
• Jupyter Notebook (for training and testing the model)

Running the Model

The machine learning model can be trained and tested using Jupyter Notebooks. The recommended crops will be shown based on the input conditions.

Team Lead:

• Dhruval Anandkar
  Lead Developer and Researcher

Team Members:

• Caleb Kaminski

• Aniket Patel

• Owen Reynolds

• Benjamin Berglund

Faculty Sponsor:

Dr. Selvanayaki Kolandapalayam Shanmugam
Assistant Professor, Computer Science Department

For any inquiries, collaboration opportunities, or feedback, feel free to contact us through the details provided below:

📧 Email: danandk2@ashland.edu 🌐 LinkedIn: https://www.linkedin.com/in/dhruvalanandkar/

We are open to discussions and collaborations in the fields of IoT, machine learning, and agri-tech!