Please use this identifier to cite or link to this item:

http://hdl.handle.net/10609/98767
Title: Modelo de aprendizaje profundo/red neuronal convolucional (CNN) para clasificación de calidad de ácidos grasos por imágenes de semillas de Helianthus annuus
Author: Vega Arias, Juan Manuel
Tutor: Vegas Lozano, Esteban
Reverter Comes, Ferran
Keywords: convolutional neural network
image classification
deep learning
Issue Date: 26-Jun-2019
Publisher: Universitat Oberta de Catalunya (UOC)
Abstract: The fatty acids content classification in seeds for their later use in industry is a long and complicated process which aims to select the different seeds that would be used with the purposes each of these seeds are of best use. These purposes are, at the same time, determined by the quality content of the fatty acids in the seed. Deep neural networks, especially convolutional neural networks, have shown a remarkable capacity for image classification and pattern abstraction in many different fields, obtaining better accuracy, and faster prediction results than those obtained by classic or human methods. In this work, we build a convolutional neural network model which can classify the sunflower seeds fatty acids quality through their images. The model was developed separating the work in two main sections. First, an experimental portion in which we collected the necessary data to build our own data set from scratch to train the neural network, and second, the analytic component in which we developed the model using the data we previously collected. This model shows a high accuracy classifying different types of sunflower (Helianthus annuus L.) seeds used for different purposes depending on their fatty acids quality content.
Language: Spanish
URI: http://hdl.handle.net/10609/98767
Appears in Collections:Bachelor thesis, research projects, etc.

Share:
Export:
Files in This Item:
File Description SizeFormat 
jvegaarTFM0619memoria.pdfMemoria de TFM3.31 MBAdobe PDFView/Open

This item is licensed under a Creative Commons License Creative Commons