A Dissertation Submitted in Partial Fulfillment of the Requirements for the Degree of
Master’s in Information and Communication Science and Engineering of the Nelson
Mandela African Institution of Science and Technology
Plant pests and diseases challenge the agricultural sector. A high-yielding crop, such as tomato
which has the potential to increase income of smallholder farmers, its production is threatened
by invasive pest called Tuta absoluta. Despite many efforts made by farmers in its
management, has continued to be a great constraint, hence calling for scholars to devise
approaches of identifying and combating it before causing great losses to farmers. This study
introduces, a deep learning based approach for the identification of the pest at early stages of
tomato growth through classification of tomato leaf images. In this study, the Convolutional
Neural Network architectures (VGG16, VGG19 and ResNet50) were trained on tomato
images dataset captured from the field containing healthy and infested tomato leaves.
Evaluation of performance for each classifier was done by considering accuracy of classifying
the tomato leaf into correct category. Experimental results showed that VGG16 attained the
highest accuracy of 91.9% in classifying tomato plant leaves into correct categories. This
model can be deployed and used to establish tool for early detection of Tuta absoluta pest