Early detection of apple leaf diseasesApple leaf disease and pests is crucial for preventing crop loss and maintaining orchard productivity. In this study, we develop a deep learningDeep learning-based classification framework using EfficientNetV2M, DenseNet121, ResNet50, and MobileNetV2, trained on a custom datasetDatasets of 1256 apple leaf images from orchards in Jammu and Kashmir. The datasetDatasets spans five classes: Alternaria, Rust, Apple Scab, Healthy, and the newly observed Apple Blotch Leaf Miner. Each modelModels was trained using transfer learningTransfer learning with data augmentation and evaluated on a stratified test set. EfficientNetV2M achieved the best performance, with 99.45% accuracy and strong generalization under real-world conditions. These results indicate that CNNConvolution Neural Networks (CNN)-based modelsModels, particularly EfficientNetV2M, can be deployed for real-time plant health monitoring. This work contributes a real-world datasetDatasets, introduces pest classification alongside diseases, and offers a scalable solution for precision agriculturePrecision agriculture.