Aplicación informática para reconocimiento de la especie camu camu (Myrciaria Dubia) a través de redes neuronales convolucionales, en Iquitos Perú, durante el año 2017
Loading...
Date
Journal Title
Journal ISSN
Volume Title
Publisher
Universidad Nacional de la Amazonía Peruana
Abstract
La presente investigacion se realizó dado la incertidumbre en el reconocimiento de
especies de la flora, un problema bastante común y nocivo para la selva amazónica,
debido a que por la poca existencia de expertos en el reconocimiento de la floresta,
trae como consecuencia la pérdida de biodiversidad, es por ello que esta tesis tuvo
como ojetivo la implementacion de un software, creado a partir del uso de técnicas
de redes neuronales Convolucionales, que permita el reconocimiento de plantas de
Camu Camu a partir de sus hojas, para lo cual la metodología utilizada fue de tipo
aplicativo tecnológica con diseño experimental, teniendo en cuenta un banco de
2800 imágenes para los procesos de entrenamiento, validación y pruebas de uso
(1400 imágenes de Camu Camu y 1400 imágenes diferentes al Camu Camu).
Logrando el 100% para el caso Sensibilidad y 97% para el caso de Especificidad,
demostrardo de esa manera la eficiencia tanto en la implementacion del uso del
software como en el uso de la red neuronal convolucional en las etapas de
entrenamieto y validación; por ultimo, se acepto la hipótesis de investigación : “El
software de reconocimiento de Camu Camu, hecho a partir de técnicas de redes
neuronales Convolucionales, será una herramienta efectiva para la identificación de
plantas de Camu Camu”.
The present investigation was carried out given the uncertainty in the recognition of species of the flora, a quite common and harmful problem for the Amazon rainforest, due to the fact that due to the little existence of experts in the recognition of the forest, it results in the loss of biodiversity, this is why this thesis was aimed at the implementation of software, created from the use of convolutional neural network techniques, which allows the recognition of Camu Camu plants from their leaves, for which the methodology used It was of technological application type with experimental design, taking into account a bank of 2800 images for the processes of training, validation and use tests (1400 images of Camu Camu and 1400 images different from Camu Camu). Achieving 100% for the Sensitivity case and 97% for the Specificity case, demonstrating in this way the efficiency in the implementation of the software use and in the use of the convolutional neuronal network in the training and validation stages; Finally, the research hypothesis was accepted: "The Camu Camu recognition software, made from convolutional neural network techniques, will be an effective tool for the identification of Camu Camu plants".
The present investigation was carried out given the uncertainty in the recognition of species of the flora, a quite common and harmful problem for the Amazon rainforest, due to the fact that due to the little existence of experts in the recognition of the forest, it results in the loss of biodiversity, this is why this thesis was aimed at the implementation of software, created from the use of convolutional neural network techniques, which allows the recognition of Camu Camu plants from their leaves, for which the methodology used It was of technological application type with experimental design, taking into account a bank of 2800 images for the processes of training, validation and use tests (1400 images of Camu Camu and 1400 images different from Camu Camu). Achieving 100% for the Sensitivity case and 97% for the Specificity case, demonstrating in this way the efficiency in the implementation of the software use and in the use of the convolutional neuronal network in the training and validation stages; Finally, the research hypothesis was accepted: "The Camu Camu recognition software, made from convolutional neural network techniques, will be an effective tool for the identification of Camu Camu plants".
Description
Keywords
Desarrollo de software, Red neuronal convolucional, Camu camu, Myrciaria dubia
Citation
Collections
Endorsement
Review
Supplemented By
Referenced By
Creative Commons license
Except where otherwised noted, this item's license is described as info:eu-repo/semantics/openAccess