Ensemble-Enhanced Deep Learning for Apple Leaf Disease Detection

This work proposes a deep learning solution based on CNNs and transfer learning to classify diseases affecting apple leaves. Since apple crops are so vulnerable to diseases like apple scab, black rot and cedar apple rust early detection is important to avoid losses and obtain quality yields. The given approach involves data preparation, data enhancement and model development using novel layers of CNN and MobileNet with ResNet50. As shown in the ensemble approach, the method of classification yielded a high test accuracy of $97 \%$ that showed a healthy sign across different disease classes. These observations suggest how the deployment of automated, real-time disease diagnosis for agriculture practices could be benefit farmers in terms of effective diagnostic capabilities. Some of the recommendations of future research work includes: The model can be optimized for mobile deployment, The use of the model can be expanded to many other plant diseases.