Improving Beech Bark Disease Classification: A Multiclass Approach with CNN-MLP Fusion

Beech Bark Disease (BBD) represents a major problem for maple and beech forests everywhere and the existing classification methods are not sufficient, hence, effective and modern solutions must be found. The approach of humans is new in the fact that it develops a hybrid machine learning model that employs Convolutional Neural Networks (CNN) and Multilayer Perceptrons (MLPs) for classifying the BBD severity into six levels. The proposed model uses the CNN ability to perform efficient feature extraction on high-dimensional images and MLP fashion for classification which leads to even greater accuracy and efficiency than the manual methods usually apply. An exhaustive set of images annotated to different classes associated with the various stages of BBD was collected, and the data preprocessing was very thorough to make the data input consistent. In the CNN-MLP approach, the CNN part was trained to detect the most relevant features, while the MLP part was responsible for classifying these features into the consequent severity levels. The model was validated using a different test set reaching an overall precision of 93.6%. Thus, it can be verified as an evident step forward over existing other classification techniques. This research is indicative of how CNNs and MLP can be put together in environmental monitoring tasks, mostly in the tree disease classification. The presented hybrid model is a useful, scalable, and precise tool for BBD control; thereby reinforcing efforts that are being directed toward the protection of forests and the preservation of the ecosystem. The scope of the model will be optimized in the future as more data are collected and the applicability of the model is extended for other ecological monitoring problems. Not only the study would add to advancement in the branch of forestry disease management, but it also depicts the marvelous, revolutionary effect of the use of machine learning technologies in the scientific field of environment.