Branching Out: A Simple 5-Level Classification of Magnolia Scale Diseases Using a Combination of CNN and DT Models

Checking and sorting Magnolia Scale diseases is very important for keeping plants healthy and making sure they get the best crop output in farming. In this work, introduce a new mix using Convolutional Neural Networks (CNN) and Decision Trees (DT). It helps to sort Magnolia Scale diseases into five agreed severity levels. The mix of CNN and DT models does well. It gets 96 out of 100 right at guessing how bad the infestation with the Magnolia Scale is. Using what’s known as transfer learning, the CNN part takes important features from an existing deep learning system. Then it adjusts these learned features to special traits of a set of Magnolia Scale images for disease identification after fine-tuning. After that, adding the results from a process called CNN into a group of choices helps make the model better. It also makes it strong and gives exact levels for how serious something is. Accuracy measurements such as precision, recall, and F1-scores above 90% across all disease severity levels show that the model is dependable and accurate in spotting different stages of illnesses. The research shows that the mixed model can be very important in disease control plans for gardening. It provides accurate and sure evaluation skills needed to find diseases early on so action can be taken quickly. This study sets the building blocks for improving ways of checking diseases and it looks promising to change disease control methods in plants outside Magnolia. This could improve plant health and make farming more sustainable without just thinking about Magnolia flowers.