Enhancing Hybrid Classification for Plant Diseases With Deep Feature Selection Based on Analytical Entropy and Statistical Method

This project aims to enhance the detection and classification of plant diseases in crops such as tomatoes, potatoes, and peppers. The system uses deep learning models like MobileNet and DenseNet for efficient feature extraction, capturing critical patterns from plant images. To improve the classification accuracy, advanced feature selection techniques are employed, including mutual information, ANOVA-based methods, standardized deviation, and entropy-based criteria. These techniques help to identify the most relevant features, filtering out noise and enhancing the overall performance of the models. After feature selection, the refined features are fed into powerful classifiers—XGBoost and Random Forest (RF)—known for their accuracy and robustness in handling complex data. The system provides a web-based interface built with HTML, Flask, and JavaScript, allowing users to log in, register, and upload plant images for disease analysis. Once an image is uploaded, the system classifies the disease and provides results, assisting farmers and agricultural experts in timely diagnosis. By combining deep learning, statistical feature selection, and advanced classifiers, this system offers an effective tool for plant disease detection and management.

Keywords:

Hybrid Classification, Plant Diseases, Feature Extraction, MobileNet, DenseNet, XGBoost, Random Forest, Feature Selection, Mutual Information, ANOVA, Entropy-based Methods, Disease Classification, Agricultural Technology, Deep Learning, Image Classification, Flask, Web Application.