Apple Leaf Diseases Detection Using Deep Learning Models: A Review

Abstract

Worldwide diseases affecting apple leaves reduce production, and early diagnosis is critical to prevent losses. Models such as CNNs, ResNet, and DenseNet can score over 95% accuracy for the detection of Apple Scab, Cedar Rust, and Marssonina Blotch. Techniques like data augmentation, transfer learning, and hybrid architectures improve the robustness and generalization of these models. Challenges exist, especially because of the controlled dataset, including high computational demands and limited scalability in real-world settings. Future work will address such issues by further extending datasets for different environments, optimizing the models for use on mobile and IoT, and incorporating precision agriculture for pro-active disease management. The review underlines the strength and weakness of deep learning in sustainable production of apples.