EFFICIENT PREPROCESSING TECHNIQUES FOR ACCURATE AND ROBUST FEATURE EXTRACTION IN DEEP LEARNING-BASED TOMATO LEAF DISEASE DETECTION: A COMPREHENSIVE REVIEW

Abstract

Early and accurate detection of tomato leaf diseases is essential for minimizing crop losses and improving agricultural productivity. Deep learning models such as Convolutional Neural Networks (CNNs), Region-Based CNNs (R-CNNs), and other advanced classifiers have demonstrated strong potential; however, their performance greatly depends on the quality of preprocessing and the robustness of the extracted features. This comprehensive review examines state- of-the-art preprocessing techniques that enhance image quality, improve feature representation, and strengthen the reliability of deep learning-based tomato leaf disease detection systems. Key methods—including image scaling, color space transformations, noise reduction, contrast enhancement, segmentation techniques such as thresholding and flood filling, as well as feature extraction approaches like gradient- based descriptors and moment-based features—are critically analyzed