Deep Learning Based Apple Detection: A Comparative Analysis of CNN Architectures

This study explores the application of deep learning techniques for the identification of apple diseases, with a particular emphasis on apple scab—a significant challenge affecting apple production in Kazakhstan. The need for automated detection solutions arises from the limitations of traditional manual inspection methods, which are labor-intensive, subjective, and time-consuming. In this research, three deep learning models—AlexNet, ResNet50, and DenseNet121—were trained and evaluated using a dataset of 3,171 images comprising both healthy and diseased apple leaves. Model development and testing were conducted using Python, TensorFlow, and Keras, with data augmentation techniques applied to enhance generalizability. Among the models tested, DenseNet121 demonstrated superior performance, achieving an accuracy of 99.35%. These results underscore the potential of deep learning in improving the efficiency and accuracy of apple disease detection, offering a scalable approach for agricultural disease management.