Publication: Reconocimiento de actividad mediante pulsera con sensores
Loading...
Official URL
Full text at PDC
Publication Date
2017
Advisors (or tutors)
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Citations
Exportar
Abstract
Desde hace unos años, con el desarrollo de los smartphones y actualmente con el auge del internet de las cosas, la tecnología ha invadido gran parte de nuestra vida como en campos de la salud, el deporte o el bienestar. Este trabajo se centra en los dos últimos aspectos con el fin de hacer una app para smartphone de reconocimiento de actividad mediante sensores, con el que se busca conseguir que las personas lleven un control de las actividades que hacen a lo largo del día, pudiendo tener un reconocedor personalizado, a diferencia de otras aplicaciones comerciales.
Se trata de un trabajo que está compuesto por una parte de aprendizaje automático, que se ha llevado a cabo gracias a los árboles de decisión que sirven para la clasificación de las actividades y una app Android que puede tomar los datos de los propios sensores del smartphone o de una pulsera Texas Instruments que se conecta mediante Bluetooth al smartphone. En la app en la que se implanta el árbol de decisión que hará la clasificación, también puedes ver tu historial de actividades. También puedes registrar nuevas actividades mediante un servidor externo a la aplicación que se encarga de recibir datos de las nuevas actividades y de generar un nuevo clasificador que incorpora esa actividad, el cual es implantado automáticamente en la app para que sea capaz de reconocer esta nueva actividad durante el uso de la app.
En este documento se describe el trabajo realizado, los sensores utilizados, las tecnologías empleadas, el proceso para la implementación de un reconocedor, el desarrollo de la app, los experimentos realizados, las conclusiones finales y por último las posibles mejoras que se pueden llevar a cabo en un futuro.
For some years, with the development of smartphones and now with the rise of the internet of things, technology has invaded much of our lives as in fields of health, sports or well-being. This work focuses on the last two aspects to make a smartphone app for recognition of activity through sensors, which seeks to get people to take control of the activities they do throughout the day, being able to have a custom recognizer, unlike other commercial applications. It is a work that is composed of a part of automatic learning, which has been carried out thanks to the decision trees that serve for the classification of activities and an Android app that can take data from the smartphone's own sensors or a Texas Instruments wristband that connects via Bluetooth to the smartphone. In the app that implements the decision tree that will do the classification, you can also see your activity history. You can also register new activities using a server external to the application that is responsible for receiving data from new activities and generating a new classifier that incorporates that activity, which is automatically implemented in the app to be able to recognize this new activity During the use of the app. This document describes the work done, the sensors used, the technologies used, the process for the implementation of a recognizer, the development of the app, the experiments carried out, the final conclusions and finally the possible improvements that can lead to in the future.
For some years, with the development of smartphones and now with the rise of the internet of things, technology has invaded much of our lives as in fields of health, sports or well-being. This work focuses on the last two aspects to make a smartphone app for recognition of activity through sensors, which seeks to get people to take control of the activities they do throughout the day, being able to have a custom recognizer, unlike other commercial applications. It is a work that is composed of a part of automatic learning, which has been carried out thanks to the decision trees that serve for the classification of activities and an Android app that can take data from the smartphone's own sensors or a Texas Instruments wristband that connects via Bluetooth to the smartphone. In the app that implements the decision tree that will do the classification, you can also see your activity history. You can also register new activities using a server external to the application that is responsible for receiving data from new activities and generating a new classifier that incorporates that activity, which is automatically implemented in the app to be able to recognize this new activity During the use of the app. This document describes the work done, the sensors used, the technologies used, the process for the implementation of a recognizer, the development of the app, the experiments carried out, the final conclusions and finally the possible improvements that can lead to in the future.
Description
Trabajo de Fin de Grado en Ingeniería Informática (Universidad Complutense, Facultad de Informática, curso 2016/2017)