Crop Prediction and Disease Detection in Cotton Using Machine Learning

Abstract

Cotton is one of the most important cash crops grown globally, and its yield is critical to the world\'s economy. Crop prediction and disease detection in cotton are essential for farmers to make informed crop management and production decisions. This project paper proposes a machine learning-based approach to predict cotton yield and detect disease in cotton plants. The proposed approach involves collecting data from various sources, including weather data, soil data, and satellite images. The collected data will be preprocessed and analyzed to identify patterns and trends. Machine learning models such as regression, classification, and clustering will be developed to predict cotton yield and detect disease in cotton plants. The proposed approach will be implemented in a web-based application that will provide farmers with timely and accurate information about the yield and health of their cotton crops. The application will also provide recommendations on crop management practices based on data analysis and predictions. The proposed project aims to provide a cost-effective and efficient solution to cotton farmers for predicting crop yield and detecting disease in cotton plants. The proposed approach can also be extended to other crops, making it applicable to a wide range of agriculture-based industries.

Introduction

I. INTRODUCTION

A. Motivation

Cotton is one of the most important cash crops grown globally, and its yield is critical to the world's economy. However, the yield of cotton crops is affected by various factors such as weather conditions, soil fertility, and disease outbreaks. Predicting cotton yield and detecting disease in cotton plants are essential for farmers to make informed crop management and production decisions.

Traditional methods of predicting cotton yield and detecting disease in cotton plants are time-consuming, labor-intensive, and inaccurate. Machine learning-based approaches can provide a cost-effective and efficient solution to this problem. This project paper proposes a machine learning-based approach to predict cotton yield and detect disease in cotton plants.

B. Problem Definition

The problem with cotton leaf disease detection is that it is currently being done manually, which can lead to errors and inconsistencies. The manual process involves inspecting each cotton leaf individually, which is time-consuming and requires a great deal of expertise.

Additionally, the accuracy of the detection process relies heavily on the subjectivity of the inspector, which can cause variations in disease identification between different individuals. As a result, farmers are often not aware of the presence of the disease until it has caused significant damage to their crops, leading to reduced yield and economic loss. A more efficient and accurate approach is needed to detect cotton leaf disease early on to prevent crop damage and loss. This problem can be addressed by developing a machine-learning model that can accurately detect cotton leaf disease in an automated and consistent manner.

II. SCOPE

The scope of the proposed project is to use Machine Learning algorithms to detect cotton leaf diseases from images. The objective is to develop a robust model that can accurately classify cotton leaves images into healthy or diseased categories. The model will leverage Convolutional Neural Network (CNN) algorithms to extract critical features from input images using several layers of convolution, pooling, and fully connected layers.

III. OBJECTIVES

1. Data Collection and Preprocessing: Collect cotton leaf images dataset from various sources and preprocess the images to remove noise and artifacts.
2. Data Augmentation and Splitting: Augment the dataset by applying transformations like random scaling, rotation, and shearing. Split the dataset into training and validation sets.
3. Model Architecture Design: Design the CNN model architecture consisting of multiple Convolutional layers, pooling layers, and fully connected layers.
4. Training and Evaluation: Train the model on the training data, and validate and optimize performance on the validation data. Evaluate model performance on the test data.
5. Deployment: Once the model is developed and validated, deploy the model as a web-based application that can take the input image and return the predicted output class of the image.
6. Future Scope: Develop an Android mobile application that integrates with the ML model to allow farmers to take images of their cotton plants and classify them into healthy or diseased categories.

The project aims to provide a cost-effective and timely solution for detecting cotton leaf diseases that can help farmers take corrective measures to control the disease before causing significant crop damage. The proposed solution can help increase crop quality, productivity, and profitability for farmers.

IV. MATERIALS AND METHODS

The proposed approach involves collecting data from various sources, including weather data, soil data, and satellite images. The collected data will be preprocessed and analysed to identify patterns and trends. Machine learning models such as regression, classification, and clustering will be developed to predict cotton yield and detect disease in cotton plants. Regression models will predict cotton yield based on weather and soil data. Classification models will be used to detect disease in cotton plants based on satellite images. Clustering models will be used to group cotton plants based on their health status. The proposed approach will be implemented in a web-based application that will provide farmers with timely and accurate information about the yield and health of their cotton crops. The application will also provide recommendations on crop management practices based on data analysis and predictions.

V. RESULTS

The proposed approach will be evaluated using real-world data collected from cotton farms. The performance of the machine learning models will be evaluated based on their accuracy, precision, recall, and F1 score. The web-based application will be evaluated based on its usability, performance, and effectiveness in providing timely and accurate information to farmers.

VI. DISCUSSION

The proposed approach provides a cost-effective and efficient solution to cotton farmers for predicting crop yield and detecting disease in cotton plants. The proposed approach can also be extended to other crops, making it applicable to a wide range of agriculture-based industries.

VII. ARCHITECTURE DIAGRAM

1. System Architecture: Smart Crop Cultivation using Logistic Regression

B. System Architecture: Disease Prediction Using Cnn

VIII. FUTURE WORK

Machine learning has been used for cotton disease detection. One work proposes a Support Vector Machine based regression system for the identification and classification of five cotton leaf diseases

i.e. Bacterial Blight, Alternaria, Gray Mildew, Cereospra, and Fusarium wilt. After disease detection, the name of a disease with its remedies will be provided to the farmers using Android app1. Another work uses Transfer learning (ResNet50) and KNN machine learning algorithms to detect cotton leaf diseases. RESNET50 distinguishes healthy and unhealthy leaves with an accuracy of 95% after training with sufficient data. KNN algorithm identifies the disease in the leaf with 86% accuracy

Conclusion

In conclusion, the proposed machine learning-based approach provides a cost-effective and efficient solution to cotton farmers for predicting crop yield and detecting disease in cotton plants. The proposed approach can be extended to other crops, making it applicable to a wide range of agriculture-based industries. The web-based application developed in this project paper provides farmers with timely and accurate information about the yield and health of their cotton crops, enabling them to make informed decisions regarding crop management.

References

[1] G. Harshitha, S. Kumar, S. Rani and A. Jain, \"Cotton disease detection based on deep learning techniques,\" 4th Smart Cities Symposium (SCS 2021), Online Conference, Bahrain, 2021, pp. 496-501, doi: 10.1049/icp.2022.0393. https://ieeexplore.ieee.org/document/9770 626 [2] S. Bavaskar, V. Ghodake, G. Deshmukh, P. Chillawar and A. Kathole, \"Image Classification Using Deep Learning Algorithms for Cotton Crop Disease Detection,\" 2022 IEEE International Conference on Distributed Computing and Electrical Circuits and Electronics (ICDCECE), Ballari, India, 2022, pp. 1-8, doi: 10.1109/ICDCECE53908.2022.9792911. https://ieeexplore.ieee.org/document/9792 911 [3] M. Meena, S. V. K and M. G, \"Plant Diseases Detection Using Deep Learning,\" 2022 1st International Conference on Computational Science and Technology (ICCST), CHENNAI, India, 2022, pp. 549-553, doi: 10.1109/ICCST55948.2022.10040431.https://ieeexplore.ieee.org/document/1004 0431

Copyright

Copyright © 2023 Mr. C.A. Ghuge , Tanmay Gangurde, Uday Narsale, Mihir Pawar, Ajinkya Popale. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.