Clasificación de especies de árboles forestales amazónicos a partir de hojas utilizando un modelo híbrido de aprendizaje automático supervisado

• Autor: Cardenas Vigo, Rodolfo
• Formato: tesis de maestría
• Fecha de Publicación: 2024
• Institución: Universidad Nacional De La Amazonía Peruana
• Repositorio: UNAPIquitos-Institucional
• Lenguaje: español
• OAI Identifier: oai:repositorio.unapiquitos.edu.pe:20.500.12737/10396
• Enlace del recurso: https://hdl.handle.net/20.500.12737/10396
• Nivel de acceso: acceso abierto
• Materia: Aprendizaje automático; Algoritmos computacionales; Clasificación; Árboles forestales; Hojas; https://purl.org/pe-repo/ocde/ford#2.02.04

The loss of biodiversity and criminal acts as a consequence of the lack of monitoring and control are closely linked to the scarce technology and limited supply of tools that support the activities of identification of the species of biodiversity present in the territory, mainly there are difficulties for the identification and classification of species of forest trees in the Amazon. For this reason, the objective of this research has been the development and implementation of a solution based on the use of Artificial Intelligence in the research line of image recognition, which has resulted in a hybrid model of supervised machine learning, which facilitates the classification of up to 40 species of Amazonian forest trees in Peru. A comparison has been made between different models of convolutional neural network (CNN) algorithms and classification algorithms such as vector machine (SVM) and logistic regression (LR), based on a set of processed leaf images of 40 species of trees of forest importance. The most outstanding results point to the hybrid model using the CNN MobileNet model and the Logistic Regression (LR) algorithm as the best solution, concluding that this model achieves high rates in the main metrics such as Accuracy, Sensitivity, Specificity and F1-score, being its average model performance 99%, which demonstrates its efficiency for this classification case, which uses leaf images to identify Amazonian forest tree species.