Publication: CIU TOUR: reconocimiento de edificios y monumentos mediante aprendizaje automático
Loading...
Official URL
Full text at PDC
Publication Date
2019
Advisors (or tutors)
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Citations
Exportar
Abstract
El sector turístico busca permanentemente nuevas propuestas para mejorar las experiencias de sus clientes. Fruto del uso de nuevas tecnologías principalmente a través de dispositivos móviles, las personas desean en la actualidad obtener una mayor información de su entorno con el menor esfuerzo posible. En el turismo, esta tendencia se plasma, entre otras propuestas, en mejorar las experiencias de los usuarios para que sean cada vez más inmersivas y cómodas. En este contexto surge este proyecto.
El objetivo principal de este proyecto ha sido el diseño e implementación de una aplicación móvil para visitas turísticas que permite al usuario obtener información sobre edificios y monumentos del entorno. Para ello, se ha realizado una captura de los requisitos necesarios para el desarrollo de la aplicación mediante un análisis de la competencia, extrayéndose requisitos funcionales a nivel de usuario y administración, y también de usabilidad. También se han estudiado las herramientas de aprendizaje automático disponibles para la clasificación de imágenes.
Ello ha permitido el uso e implementación de una red neuronal para la clasificación de imágenes. El resultado es una aplicación capaz de obtener de manera sencilla una imagen del entorno, clasificarla en una categoría (que corresponde a un edificio o monumento) y mostrar información relacionada con ésta. De este modo, el usuario puede realizar su propio tour virtual usando la información del proveedor.
La aplicación se ha evaluado desde el punto de vista de la clasificación y reconocimiento de imágenes y de la experiencia del usuario. Se ha concluido que la aplicación ha satisfecho en general a los usuarios en cuanto a la funcionalidad, utilidad, comodidad e innovación, a pesar de posibles correcciones o mejoras que puedan realizarse. Cabe destacar que, tras analizar el proceso de clasificación, se detectó la necesidad de realizar una optimización de dicho proceso mediante la ampliación del número y variedad del conjunto de imágenes usadas para el entrenamiento de la red neuronal y el uso de herramientas de optimización ofrecidas por TensorFlow.
Como mejoras futuras de la aplicación, se plantea la optimización del proceso de clasificación, la implementación de realidad virtual y/o realidad aumentada, y mejoras en la interfaz de usuario.
The tourism sector is constantly looking for new proposals to improve the experiences of its customers. As a result of the use of new technologies, mainly through mobile devices, people currently want to obtain more information about their environment with the least possible effort. In tourism, this trend is reflected, among other proposals, in improving the experiences of users to become increasingly immersive and comfortable. It is in this context that this project arises. The main objective of this project has been the design and implementation of a mobile application for tourists that allows the user to obtain information on buildings and monuments in the environment. To this end, a capture of the necessary requirements for the development of the application has been carried out through an analysis of the competition, extracting functional requirements at user and administration level, as well as usability. The automatic learning tools available for image classification have also been studied. This has allowed the use and implementation of a neural network for image classification. The result is an application capable of easily obtaining an image of the environment, classifying it into a category (corresponding to a building or monument) and displaying information related to it. In this way, the user can make his own virtual tour using the information of the provider. The application has been evaluated from the point of view of the classification and recognition of images and the user experience. It has been concluded that the application has generally satisfied users in terms of functionality, usefulness, comfort and innovation, despite possible corrections or improvements that may be made. It should be noted that after analyzing the classification process, it was detected the need to perform an optimization of this process by increasing the number and variety of images used for training the neural network and the use of optimization tools offered by TensorFlow. Future improvements to the application include the optimization of the classification process, the implementation of virtual reality and/or augmented reality, and improvements to the user interface.
The tourism sector is constantly looking for new proposals to improve the experiences of its customers. As a result of the use of new technologies, mainly through mobile devices, people currently want to obtain more information about their environment with the least possible effort. In tourism, this trend is reflected, among other proposals, in improving the experiences of users to become increasingly immersive and comfortable. It is in this context that this project arises. The main objective of this project has been the design and implementation of a mobile application for tourists that allows the user to obtain information on buildings and monuments in the environment. To this end, a capture of the necessary requirements for the development of the application has been carried out through an analysis of the competition, extracting functional requirements at user and administration level, as well as usability. The automatic learning tools available for image classification have also been studied. This has allowed the use and implementation of a neural network for image classification. The result is an application capable of easily obtaining an image of the environment, classifying it into a category (corresponding to a building or monument) and displaying information related to it. In this way, the user can make his own virtual tour using the information of the provider. The application has been evaluated from the point of view of the classification and recognition of images and the user experience. It has been concluded that the application has generally satisfied users in terms of functionality, usefulness, comfort and innovation, despite possible corrections or improvements that may be made. It should be noted that after analyzing the classification process, it was detected the need to perform an optimization of this process by increasing the number and variety of images used for training the neural network and the use of optimization tools offered by TensorFlow. Future improvements to the application include the optimization of the classification process, the implementation of virtual reality and/or augmented reality, and improvements to the user interface.
Description
Trabajo de Fin de Grado, Universidad Complutense, Facultad de Informática, Departamento de Ingeniería del Software e Inteligencia Artificial. Curso 2018/2019