Authors :
Nahid Hasan Ashik; Rakibul Islam; Md. Arif Hossain; Toukir Ahammed; Md Shakirul Islam
Volume/Issue :
Volume 8 - 2023, Issue 10 - October
Google Scholar :
https://tinyurl.com/bdfx23wh
Scribd :
https://tinyurl.com/5222ap3e
DOI :
https://doi.org/10.5281/zenodo.10029575
Abstract :
In a country where agriculture is the primary
industry, plant diseases can have a wide range of adverse
effects on the economy and the management of food
resources. The increasing occurrence of plant diseases
significantly threatens global food security and plant
productivity. Classification is constantly constrained by
issues like overfitting and low accuracy as potential
diseases manifest on plant leaves. The agricultural sector
needs accurate and error-free analysis to distinguish
healthy products from defective ones. An effective model
of autonomous feature extraction that has been
demonstrated to be reasonably effective for detection
and classification tasks is deep convolutional neural
networks. Deep convolutional neural networks, on the
other hand, frequently require a substantial amount of
training data, cannot be translated, and require a
number of parameters to be specified and adjusted. In
this study, we suggested a hybrid deep learning
approach for quickly identifying and categorizing
different plant leaf diseases. This hybrid system
combines support vector machines (SVM), convolutional
neural networks (CNN), and both.
To summarize, we extract features through model
engineering (ME). To improve feature discrimination
and processing speed, support vector machine (SVM)
models are used. Using the datasets, 25 leaf image sets of
healthy and diseased leaves of tomato, potato, grape,
apple, and corn were analyzed. There were roughly
31397 images produced as a result of the classification
process. SVM is a key component of hybrid feature
selection; this algorithm's goal is to produce the desired
output with the fewest features possible. By utilizing the
aforementioned evaluation criteria, a comparative
analysis of both techniques is presented.
Keywords :
Plant Disease Detection, Image Processing, CNN, VGG19, Xception, DenseNet201, ResNet15V2, SVM.
In a country where agriculture is the primary
industry, plant diseases can have a wide range of adverse
effects on the economy and the management of food
resources. The increasing occurrence of plant diseases
significantly threatens global food security and plant
productivity. Classification is constantly constrained by
issues like overfitting and low accuracy as potential
diseases manifest on plant leaves. The agricultural sector
needs accurate and error-free analysis to distinguish
healthy products from defective ones. An effective model
of autonomous feature extraction that has been
demonstrated to be reasonably effective for detection
and classification tasks is deep convolutional neural
networks. Deep convolutional neural networks, on the
other hand, frequently require a substantial amount of
training data, cannot be translated, and require a
number of parameters to be specified and adjusted. In
this study, we suggested a hybrid deep learning
approach for quickly identifying and categorizing
different plant leaf diseases. This hybrid system
combines support vector machines (SVM), convolutional
neural networks (CNN), and both.
To summarize, we extract features through model
engineering (ME). To improve feature discrimination
and processing speed, support vector machine (SVM)
models are used. Using the datasets, 25 leaf image sets of
healthy and diseased leaves of tomato, potato, grape,
apple, and corn were analyzed. There were roughly
31397 images produced as a result of the classification
process. SVM is a key component of hybrid feature
selection; this algorithm's goal is to produce the desired
output with the fewest features possible. By utilizing the
aforementioned evaluation criteria, a comparative
analysis of both techniques is presented.
Keywords :
Plant Disease Detection, Image Processing, CNN, VGG19, Xception, DenseNet201, ResNet15V2, SVM.
