Tea agriculture is a very crucial part of the agricultural economy to which proper identification of leaf diseases is relevant when confirming the quality and sustainability of yield. This paper will discuss how good the classifiers can be on examining specific leaf diseases of tea using images, using deep learning models. The four state-of-the-art convolutional neural network (CNN) models, which consist of DenseNet201, ResNet152V2, VGG19, and Xception, were used in comparing them. DenseNet201 obtained the best performance with an accuracy of (94.03%) surpassing all other networks namely ResNet152V2 (80.12%), Xception (72%), and VGG19 (80.43%) in novel network architecture task. The findings prove that deep CNN architectures can be an effective tool in enabling smart farming and precision plant health diagnosis. The research can offer practical information towards using a model when deciding where to implement AI-enabled plant disease, especially in the tea sector.