Tomato Leaf Disease Classification Using Mixed Pooling with Nesterov Accelerated Gradient-Based Convolutional Neural Network

Abstract

Tomato leaf disease classification plays an important part in improving agricultural efficiency by enabling early and accurate detection of plant diseases. Plant diseases have a crucial effect on leaves, with every disease displaying specific spots characterized by individual colors and locations. The precise classification of tomato plant disease plays a vital part in identifying the rate as well as production quality. Deep learning (DL) has arisen as an important approach for determining plant diseases because of its remarkable performance. Thus, this research proposes the Mixed Pooling with Nesterov Accelerated Gradient-based Convolutional Neural Network (MPNAGCNN) for accurate multi-class classification of tomato leaf diseases. The proposed model integrates mixed pooling to preserve both dominant and subtle disease features, along with the NAG optimizer to achieve faster and more stable convergence. Primarily, this research considers two different benchmark datasets, such as PlantVillage and tomato leaf disease, for evaluating the proposed method. In addition, the different preprocessing approaches, named image resizing and data augmentation, are performed to enhance the classification performance. The experimental results demonstrate that the proposed MPNAGCNN approach reaches better accuracy of 99.95% and 99.98% as compared to the existing methods such as Capsule Neural Network (CapsNet) and DenseNet121. The findings illustrate that the proposed method provides lightweight, effective, and more precise problemsolving for automated tomato leaf disease detection.