Apple leaf diseases must be identified and prevented as soon as possible in order to improve productivity. Since convolutional neural networks (CNN) have produced significant achievements in the field of machine vision, deep CNN models are used in this work to identify and diagnose diseases in apples from their leaves. When CNN models are developed from the scratch, they require many parameters and a significant computational cost. In this work, we therefore used standard CNN pre trained networks albeit with optimisation to speed up the training and reduce computational cost. The implemented models were trained with an open dataset consisting of 3 different classes of apple diseases and one healthy class. The implemented models achieved a disease-classification accuracy rates of 98.13% and 96.12%, using ResNet1 and InceptionV3, which were greater than that of traditional handcraftedfeature-based approaches. In comparison with other deep-learning models, the implemented model achieved better performance in terms of accuracy, and it required less training time.