Comedero de aves inteligente: sistema de clasificación, monitoreo y reporte de especies de aves por medio de un comedero construido con un ordenador de placa reducida y análisis de imágenes usando Machine Learning.
Fecha
2019-04
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad de Guayaquil Facultad de Ciencias Matemáticas y Físicas Carrera de Ingeniería en Sistemas Computacionales
Resumen
La clasificación de imágenes mediante inteligencia artificial es un tema que está dando
mucho de qué hablar en el mundo tecnológico. Las principales empresas tecnológicas como
Facebook, Apple y Google están investigando exhaustivamente en esta área para brindar
servicios como búsqueda visual, etiquetado automático de amigos en publicaciones de
redes sociales y la capacidad de usar tu rostro para desbloquear tu teléfono celular. El
presente proyecto de titulación nace a partir de la idea de aplicar una las actuales técnicas
de aprendizaje automatizado como lo son las redes neuronales convolucionales que es un
algoritmo matemático que puede interpretar y aprender de características y patrones
incrustados en imágenes, este potente algoritmo se lo utilizará para crear una herramienta
tecnológica de apoyo para observadores de aves silvestres, tanto expertos como
principiantes, la finalidad de este proyecto tiene como objetivo clasificar especies de aves
silvestres de manera automática través de un comedero para aves. Para la realización de
este proyecto se utilizarán herramientas software de uso libre como lo TensorFlow que es
marco de trabajo especializado en aprendizaje automatizado y que nos permite crear un sin
número de algoritmos para aprendizaje automático, entre ellos las redes neuronales
convolucionales que es uno de los más avanzados en el campo de aprendizaje automatizado
basado en imágenes. Además, para el almacenamiento de las imágenes que capture y
clasifique nuestro modelo de red neuronal convolucional a través de comedero se utilizará
Firebase que es una plataforma alojada en la nube que proporciona almacenamiento y
sincronización automática de datos. Para una mejor experiencia del usuario final se
concluirá con el desarrollo de una aplicación móvil para dispositivos Android donde se
visualizarán las imágenes de las aves que haya detectado e identificado nuestra red neuronal
convolucional a través del comedero.
The classification of images by artificial intelligence is a topic that is giving much to talk about in the technological world. Leading technology companies such as Facebook, Apple, and Google are doing extensive research in this area to provide services such as visual search, automatic tagging of friends in social networking publications, and the ability to use your face to unlock your cell phone. The present degree project is born from the idea of applying one the actual automated learning techniques such as convolutional neural networks which is a mathematical algorithm that can interpret and learn from features and patterns embedded in images, this powerful algorithm will be used to create a technological supportive tool for wild birds watchers, both experts or beginners, the purpose of this project is to identify automatically identify wild bird species through a bird feeder. In order to carry out this project, free software tools will be used, such as the TensorFlow, which is a specialized framework for automated learning that allows us to create a number of algorithms for machine learning, including convolutional neural networks. one of the most advanced in the field of automated image-based learning. In addition, for the storage of the images that are captured and identified by our convolutional neural network model through a bird feeder, Firebase will be used, which is a platform hosted in the cloud that provides storage services and automatic data synchronization. For a better end-user experience, we’ll conclude with the development of a mobile application for Android devices where the images of the birds that have been detected and identified by our convolutional neuronal network through the bird feeder will be displayed. TensorFlow, machine learning, convolutional neural networks, algorithms, free software.
The classification of images by artificial intelligence is a topic that is giving much to talk about in the technological world. Leading technology companies such as Facebook, Apple, and Google are doing extensive research in this area to provide services such as visual search, automatic tagging of friends in social networking publications, and the ability to use your face to unlock your cell phone. The present degree project is born from the idea of applying one the actual automated learning techniques such as convolutional neural networks which is a mathematical algorithm that can interpret and learn from features and patterns embedded in images, this powerful algorithm will be used to create a technological supportive tool for wild birds watchers, both experts or beginners, the purpose of this project is to identify automatically identify wild bird species through a bird feeder. In order to carry out this project, free software tools will be used, such as the TensorFlow, which is a specialized framework for automated learning that allows us to create a number of algorithms for machine learning, including convolutional neural networks. one of the most advanced in the field of automated image-based learning. In addition, for the storage of the images that are captured and identified by our convolutional neural network model through a bird feeder, Firebase will be used, which is a platform hosted in the cloud that provides storage services and automatic data synchronization. For a better end-user experience, we’ll conclude with the development of a mobile application for Android devices where the images of the birds that have been detected and identified by our convolutional neuronal network through the bird feeder will be displayed. TensorFlow, machine learning, convolutional neural networks, algorithms, free software.
Descripción
PDF
Palabras clave
TensorFlow, Aprendizaje automático, Redes neuronales convolucionales, Algoritmos, Software libre., TensorFlow, Machine learning, convolutional neural networks, Algorithms, Free software.