Disease Classification in Eggplant Using Pre-trained VGG16 and MSVM.

Area of Science:

• Agricultural Science
• Computer Vision
• Machine Learning

Background:

• Deep learning for crop disease classification requires extensive image datasets.
• Eggplant (Solanum melongena) is a vital crop susceptible to diseases, yet lacks a dedicated image dataset.
• Existing research highlights the need for standardized datasets to advance disease identification technologies.

Purpose of the Study:

• To develop a comprehensive image dataset for five major eggplant diseases under laboratory and field conditions.
• To evaluate the performance of deep learning models, specifically VGG16, for eggplant disease classification.
• To explore the utility of different color spaces and feature extraction methods for improved classification accuracy.

Main Methods:

• A novel image dataset was curated for five common eggplant diseases.
• Pre-trained Visual Geometry Group 16 (VGG16) architecture was employed for classification tasks.
• Images were processed in various color spaces (RGB, HSV, YCbCr, grayscale) and VGG16 features were extracted for Multi-Class Support Vector Machine (MSVM) analysis.

Main Results:

• The developed dataset, when utilized with RGB and YCbCr images under field conditions, yielded a high classification accuracy of 99.4%.
• VGG16 demonstrated strong performance, with feature extraction and MSVM achieving comparable or superior accuracy in some cases.
• Comparative analysis with other architectures confirmed the robustness of the proposed approach.

Conclusions:

• The created eggplant disease dataset is a valuable resource for advancing deep learning-based crop disease classification.
• The study validates the effectiveness of the VGG16 architecture and specific color spaces for accurate disease identification.
• Further research directions include exploring inter-class accuracy variations and optimizing feature extraction techniques.