Tales of the Rosehip: A Journey Through Neural Networks and Whispering Forests

Infectious plant diseases are a major hindrance to agricultural output and viability. Effective disease management relies on correct and prompt disease diagnosis. For the purpose of multi-class categorization of six frequent diseases affecting rosehip leaves, the authors of this study propose a hybrid model that incorporates Convolutional Neural Networks and Random Forests. There are four convolutional layers, one max-pooling layer, and then a Random Forest classifier in the model's architecture. Rosehip leaf photos were annotated with information about six different diseases (Diplocarpon rosae, Podosphaera pannosa, Phragmidium, Rose Mosaic Disease, Cercospora rosicola, and Rose Rosette Disease; RRD) to assess the model's accuracy. The model performed exceptionally well, with high precision, recall, and F1 scores for every disease category. Micro and macro average scores both showed that the model could correctly detect diseases with a 99% success rate. Their results corroborate the usefulness of a hybrid strategy, in which the CNN extracts useful characteristics from the input photos and the Random Forest classifier makes good use of them to accurately categorize diseases. The model's promising results suggest it might be used in real-world settings, such as precision agriculture and automated plant disease diagnosis. The successful application of the suggested hybrid model to the identification of rosehip leaf diseases advances the creation of novel and dependable disease detection methods for agricultural systems. This study demonstrates the potential of merging deep learning and ensemble learning methods, providing a useful resource for farmers and agronomists working to improve disease management strategies and maintain rosehip plants' health and productivity.