Aplicación del aprendizaje profundo al reconocimiento de la actividad humana
Year: 2020
Type: dissertation
Abstract: This paper aims to apply deep learning techniques for the recognition of human activity. Four \nneural network models have been defined and implemented and have been trained with the \ndata obtained from wearable sensors, placed on people who were carrying out different \nphysical activities. The training dataset contains realistic information, taking into account the \nconcept of gradual displacement. After evaluating the proposed network models, it is observed \nthat the recurrent networks, by better managing the time series of data, have greater precision \nin recognizing the activities. On the other hand, convolutional networks have been shown to \nimprove the extraction of characteristics and, by establishing hybrid models, can enhance the \nexpressive capacity of recurrent networks.
Author: Óscar Piqueras-Segura
Language: Spanish
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W3103396426
