Classifying Tomato Leaf Diseases Using Diverse Deep Learning Architectures: AlexNet, DenseNet, Inception, Xception

Deep learning techniques have significantly evolved in the agricultural domain that provide ways to identify diseases in plants. Considering, the wide use of tomato crop across the globe, the identification of disease in tomato leaf for prevention of massive crop damage is a significant area of study. Thus, this work aims to identify diseased leaf in Tomato plant. The use of deep learning techniques help to better classify the disease as better quality features are extracted. The experiments are performed using four deep neural-network architecture models viz. AlexNet, DenseNet, Inception and Xception on Tomato Leaf Disease dataset. In this process, firstly, t he r aw i mages are preprocessed and segmented which are subsequently fed to the respective four models. Each model extracts the features based on their architecture and produces the classification results with train-test-validation mechanism. Among all models, Exception and AlexNet models performs better than others in terms of accuracy metric having values as 95% and 90% respectively.