NORMA eResearch @NCI Library

Detection and Classification of Leaf Diseases in Maize Plant using Machine Learning

Jayakumar, Adarsh (2020) Detection and Classification of Leaf Diseases in Maize Plant using Machine Learning. Masters thesis, Dublin, National College of Ireland.

[thumbnail of Master of Science]
PDF (Master of Science)
Download (1MB) | Preview
[thumbnail of Configuration manual]
PDF (Configuration manual)
Download (783kB) | Preview


One of the major challenges faced by agricultural industry is the need for accurate and early detection of diseases that affect crops. Diseases affect the quality of crops and are capable of wiping out hectares of crop yield resulting in major loss to farmers. Current diagnostic techniques are time consuming and require the presence of highly skilled professionals to analyze the affected plants, understand the symptoms, identify the disease and thereby suggest suitable remedies. The limitations of such techniques have enforced the need to look for alternative techniques which can detect and classify diseases at an early stage. Smart farming using suitable infrastructure can help in tackling and providing solutions to such problems. Data mining techniques, in the recent years have shown great promise in identifying and classifying patterns in similar areas of research. Current research aims to evaluate the performance of algorithms like XGBoost, Gradient Boost, Convolutional Neural Network(CNN) and its architectures like VGG16 and VGG19 coupled with data augmentation and transfer learning against traditional machine learning algorithms like Support Vector Machine(SVM), random forest and measure their effectiveness in identifying and classifying maize plant diseases in terms of accuracy, precision, recall and training time. Training of the models were performed on an open source database containing close to 4000 images, encompassing four distinct classes, including healthy plant images. Out of the models developed, VGG19 architecture of CNN with transfer learning performed the best by achieving an overall accuracy of 95 percent, thereby satisfying the need of building an effective and robust classification model. Also, the performance of the models developed were found to improve with increase in amount of training data. The results obtained using transfer learning techniques on CNN architectures are highly promising and can be extended further to form a comprehensive plant disease identification system that is capable of operating in real world environment. It can thus empower the agricultural community to diagnose diseases and initiate timely treatment without the intervention of trained experts.

Item Type: Thesis (Masters)
Subjects: Q Science > QA Mathematics > Electronic computers. Computer science
T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science
Q Science > QA Mathematics > Computer software
T Technology > T Technology (General) > Information Technology > Computer software
H Social Sciences > HD Industries. Land use. Labor > Specific Industries > Agriculture Industry
S Agriculture > S Agriculture (General) > Farming Industry
Divisions: School of Computing > Master of Science in Data Analytics
Depositing User: Dan English
Date Deposited: 11 Jun 2020 13:28
Last Modified: 11 Jun 2020 13:28

Actions (login required)

View Item View Item