Use Of Convolutional Neural Networks In Detection Of Branch Dieback Disease In Pequi Trees

This study aims to evaluate and compare convolutional neural network (CNN) architectures, specifically AlexNet, Inception, LeNet, VGG16, and MobileNet, for diagnosing branch dieback (seca-dos-ponteiros) in pequi trees (Caryocar brasiliense). This disease represents a severe threat to tree development and fruit yield, causing substantial economic and ecological losses. The methodology implements computer vision techniques using Python libraries, including Scikit-learn, NumPy, TensorFlow, and Matplotlib, to process and evaluate visual data. The CNN models analyze images of leaves, branches, and flower spikes to detect pathological patterns, such as alterations in texture and coloration. Ultimately, this comparative analysis identifies the most effective architecture for accurately distinguishing healthy plants from diseased ones.