Classification of Angular Leaf Spot, Bean Rust Diseases in Haricot Bean Leaves using CNN and Transfer Learning

Haricot beans are a highly valuable legume that is increasingly in demand on the market. It is an important crop for Ethiopia in terms of earning foreign currency. The classification of haricot bean leaf diseases is done manually in Ethiopia by experts. It requires a lot of effort, takes a lot of time, and has accuracy and consistency issues. The purpose of this study is to minimize the effort to classify the disease accurately using deep learning and image processing. The required images of haricot beans were taken in Wentu Bereha, and Kork in the Ethiopian Amhara region’s Baso Liben Woreda. 2400 sample images of healthy haricot beans, angular leaf spots, and bean rust were collected. A smartphone camera with a 4080x3060 pixel resolution was used to get images. In addition, various classifiers like CNN, VGG, MobileNet, ResNet50, and GoogleNet were used for classifying according to their predefined classes. In this study, the Python programming language was employed to conduct the experiments. In this study, six groups of experiments applying various algorithms (CNN) and transfer learning methods (VGG16, VGG19, ResNet50, ResNet50V2, GoogleNet/InceptionV3& MobileNetV2) were carried out. The experiment’s findings showed that the CNN classifier from scratch had a testing accuracy of 87.5%, training accuracy of 88.04%, and validation accuracy of 89.57%. Furthermore, testing accuracy for the VGG16, VGG19, ResNet50, ResNet50V2, InceptionV3, and MobileNetV2 classifiers that use transfer learning techniques was shown to be 90.2%, 91.08%, 92.86%, 58.1%, 47%, and 99.9%, respectively. The model’s accuracy increased or decreased when transfer learning classifiers were used. It was determined that the model can classify the diseases of haricot bean leaf with the best percentage of accuracy in some applied techniques especially in MobileNetV2. It is advised that the developed strategy be applied to additional varieties of haricot beans, such as white beans and speckled beans.

Authors

Institutions

Publication Details

Published
2024-11-18
DOI
https://doi.org/10.1109/ict4da62874.2024.10777237
Primary Topic
Smart Agriculture and AI
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

Classification of Angular Leaf Spot, Bean Rust Diseases in Haricot Bean Leaves using CNN and Transfer Learning

Kebad Nigussie Werku, Ibrahim Gashaw Kassa, Abinet Bizuayehu Desta, Dessalegn Melkie Ayal et al.
Smart Agriculture and AI
article

Classification of Angular Leaf Spot, Bean Rust Diseases in Haricot Bean Leaves using CNN and Transfer Learning

Kebad Nigussie Werku, Ibrahim Gashaw Kassa, Abinet Bizuayehu Desta, Dessalegn Melkie Ayal, Mesenbet Abewa Bimrew, Misganaw Ketema Ayele
article en

Abstract

Haricot beans are a highly valuable legume that is increasingly in demand on the market. It is an important crop for Ethiopia in terms of earning foreign currency. The classification of haricot bean leaf diseases is done manually in Ethiopia by experts. It requires a lot of effort, takes a lot of time, and has accuracy and consistency issues. The purpose of this study is to minimize the effort to classify the disease accurately using deep learning and image processing. The required images of haricot beans were taken in Wentu Bereha, and Kork in the Ethiopian Amhara region’s Baso Liben Woreda. 2400 sample images of healthy haricot beans, angular leaf spots, and bean rust were collected. A smartphone camera with a 4080x3060 pixel resolution was used to get images. In addition, various classifiers like CNN, VGG, MobileNet, ResNet50, and GoogleNet were used for classifying according to their predefined classes. In this study, the Python programming language was employed to conduct the experiments. In this study, six groups of experiments applying various algorithms (CNN) and transfer learning methods (VGG16, VGG19, ResNet50, ResNet50V2, GoogleNet/InceptionV3& MobileNetV2) were carried out. The experiment’s findings showed that the CNN classifier from scratch had a testing accuracy of 87.5%, training accuracy of 88.04%, and validation accuracy of 89.57%. Furthermore, testing accuracy for the VGG16, VGG19, ResNet50, ResNet50V2, InceptionV3, and MobileNetV2 classifiers that use transfer learning techniques was shown to be 90.2%, 91.08%, 92.86%, 58.1%, 47%, and 99.9%, respectively. The model’s accuracy increased or decreased when transfer learning classifiers were used. It was determined that the model can classify the diseases of haricot bean leaf with the best percentage of accuracy in some applied techniques especially in MobileNetV2. It is advised that the developed strategy be applied to additional varieties of haricot beans, such as white beans and speckled beans.

Debark University (ET), University of Gondar (ET)
Zero hunger
Openalex Percentile: Top 87%
Smart Agriculture and AI
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.