Analysis of Different CNN Architectures for Tomato Leaf Disease Classification
Mamta Gehlot, Madan Lal Saini
Abstract
In the field of agriculture disease in crops are major problem. The main aim of the work is to classify disease in tomato leaf using different Convolution Neural Network architectures. The disease in tomato leaf can affect quality and quantity of production. To overcome this problem early disease identification, classification and detection is required. Recently, deep learning is very popular object recognition and detection. Convolution Neural Network id part of deep learning which is widely used in object detection part. In these different architectures of Convolution Neural Network are used to classify 10 classes of tomato disease. This paper used AlexNet, VGG-16, GoogleNet, DenseNet-121, and ResNet-lOl for classification of tomato leaves. The dataset used is publicly available PlantVillage dataset. DenseNet-121, VGG16 and ResNet-lOl performs almost equal in terms of accuracy, precision, Fl-score, and Recall. Size of DenseNet-121 is only 89.6MB which is very smaller than ResNet-lOl and all other models.