Plant disease severity classification using deep learning

Over the years, there are improvements in food accessibility and production. However, food security is a major concern in which various factors, such as plant diseases, threaten it. Plant diseases cause significant damage to the crops and substantially reduce food production. In the past, early dete...

Full description

Saved in:
Bibliographic Details
Main Author: Lee, San Kong
Format: Thesis
Language:English
Published: 2020
Subjects:
Online Access:http://eprints.utm.my/id/eprint/93102/1/LeeSanKongMSKE2020.pdf
Tags: Add Tag
No Tags, Be the first to tag this record!
id my-utm-ep.93102
record_format uketd_dc
spelling my-utm-ep.931022021-11-07T06:00:42Z Plant disease severity classification using deep learning 2020 Lee, San Kong TK Electrical engineering. Electronics Nuclear engineering Over the years, there are improvements in food accessibility and production. However, food security is a major concern in which various factors, such as plant diseases, threaten it. Plant diseases cause significant damage to the crops and substantially reduce food production. In the past, early detection of plant diseases is done by experts equipped with academic knowledge background and practical experience on plant symptoms. This process is complicated and time-consuming when it comes to the classification task of plant disease with a limited resource of knowledge and plant experts. Also, the current image analysis has a limitation of localizing and classifying diseases with almost the same and similar symptoms, such as early blight and late blight. This will impact the best treatment time for the plant before the disease spreading out. Hence, the objective of this work is to develop a high accuracy deep learning model that can classify early blight and late blight disease into low, mild, and severe levels based on transfer learning with ResNet. The process involves in this project is collecting leaf images of the diseased and healthy plant from the Plant Village dataset and classifying them into low, mild, and severe folders. With the dataset ready, the deep learning model is trained based on the transfer learning with ResNet. The model’s parameters and training settings will be varied and optimized to enhance the accuracy of the model. Performance comparison will be made before and after the optimization. Based on the results, the best accuracy that has been archived by plant disease severity level classification system is 80.01%. Also, the recall rate achived by the system for low, mild and severe are 89.55%, 100%, and 76.32% respectively. This project will ease the process of classifying the severity level of the plant disease, which reduces the damage on the crops due to the early detection of the diseases. 2020 Thesis http://eprints.utm.my/id/eprint/93102/ http://eprints.utm.my/id/eprint/93102/1/LeeSanKongMSKE2020.pdf application/pdf en public http://dms.library.utm.my:8080/vital/access/manager/Repository/vital:135933 masters Universiti Teknologi Malaysia, Faculty of Engineering - School of Electrical Engineering Faculty of Engineering - School of Electrical Engineering
institution Universiti Teknologi Malaysia
collection UTM Institutional Repository
language English
topic TK Electrical engineering
Electronics Nuclear engineering
spellingShingle TK Electrical engineering
Electronics Nuclear engineering
Lee, San Kong
Plant disease severity classification using deep learning
description Over the years, there are improvements in food accessibility and production. However, food security is a major concern in which various factors, such as plant diseases, threaten it. Plant diseases cause significant damage to the crops and substantially reduce food production. In the past, early detection of plant diseases is done by experts equipped with academic knowledge background and practical experience on plant symptoms. This process is complicated and time-consuming when it comes to the classification task of plant disease with a limited resource of knowledge and plant experts. Also, the current image analysis has a limitation of localizing and classifying diseases with almost the same and similar symptoms, such as early blight and late blight. This will impact the best treatment time for the plant before the disease spreading out. Hence, the objective of this work is to develop a high accuracy deep learning model that can classify early blight and late blight disease into low, mild, and severe levels based on transfer learning with ResNet. The process involves in this project is collecting leaf images of the diseased and healthy plant from the Plant Village dataset and classifying them into low, mild, and severe folders. With the dataset ready, the deep learning model is trained based on the transfer learning with ResNet. The model’s parameters and training settings will be varied and optimized to enhance the accuracy of the model. Performance comparison will be made before and after the optimization. Based on the results, the best accuracy that has been archived by plant disease severity level classification system is 80.01%. Also, the recall rate achived by the system for low, mild and severe are 89.55%, 100%, and 76.32% respectively. This project will ease the process of classifying the severity level of the plant disease, which reduces the damage on the crops due to the early detection of the diseases.
format Thesis
qualification_level Master's degree
author Lee, San Kong
author_facet Lee, San Kong
author_sort Lee, San Kong
title Plant disease severity classification using deep learning
title_short Plant disease severity classification using deep learning
title_full Plant disease severity classification using deep learning
title_fullStr Plant disease severity classification using deep learning
title_full_unstemmed Plant disease severity classification using deep learning
title_sort plant disease severity classification using deep learning
granting_institution Universiti Teknologi Malaysia, Faculty of Engineering - School of Electrical Engineering
granting_department Faculty of Engineering - School of Electrical Engineering
publishDate 2020
url http://eprints.utm.my/id/eprint/93102/1/LeeSanKongMSKE2020.pdf
_version_ 1747818633381281792