Automated Disease Detection And Classification Of Solanaceae Family Leaves: Focus On Tomato, Brinjal, And Pepper

Abstract

Plant diseases are a major concern for global agriculture, impacting a wide range of crops essential for food security. These diseases, which can be produced by bacteria, viruses, fungi, and environmental stress, lead to visible symptoms. Traditional disease detection methods lead to delayed intervention and crop damage. To address this challenge, there is a growing need for automated and efficient systems that can detect and classify plant diseases at an early stage. This research objectives to develop a Deep Learning (DL)-based finding and organization model for Solanaceae family leaves using a novel Stochastic Fractal Search-driven Intelligent Faster Region-based Convolutional Neural Network (SFS-InT-Faster R-CNN). The proposed model combines advanced techniques to optimize Faster R-CNN, ensuring improved accuracy with efficiency in detecting diseases. The dataset for this research includes high-resolution leaf images of tomato, brinjal, and pepper. The data pre-processing phase involves noise reduction using the Non-Local Means (NLM) filter and contrast enhancement via Adaptive Histogram Equalization (AHE), both crucial for improving image quality and highlighting subtle disease signs. Feature extraction is conducted through the Gray-Level Co-occurrence Matrix (GLCM), which captures critical texture and spatial features for accurate disease classification. The SFS-InT-Faster R-CNN model utilizes an SFS optimization algorithm to enhance the Faster R-CNN architecture's hyperparameters, significantly improving detection performance. The suggested model is implemented in Python software. The outcomes show that the model surpasses conventional techniques in precision of 95.2%, an accuracy of 98.5%, an F1-score of 98%, a recall of 98.36%, providing an efficient method to identify diseases in Solanaceae crops early. This research concludes that the proposed framework provides a reliable tool for plant disease monitoring, which can help optimize agricultural practices and mitigate crop losses.