Leveraging VGG19 for accurate detection of apple plant diseases

Early detection of apple leaf diseases is vital for improving crop productivity and supporting sustainable agriculture. This study proposes an enhanced VGG19-based deep learning model to classify apple foliar diseases with high precision. The model was trained on the publicly available Plant Village dataset, comprising 11,300 RGB images categorized into four classes: Apple Scab, Black Rot, Cedar Apple Rust, and Healthy. Using a holdout validation strategy with a 70:20:10 train-validation-test split, the proposed VGG19 model achieved a training accuracy of 97.50%, validation accuracy of 98.80%, and a minimal loss of 0.0389. Comparative analysis with CNN, VGG16, and InceptionV3 models demonstrated that VGG19 consistently outperforms them across all metrics. It achieved the highest precision (98.56%), recall (97.95%), and F-score (98.80%), reflecting superior classification performance. The model also showed faster convergence and lower loss across 100 epochs.