A soft computing approach to violence detection in social media for smart cities

Please use this identifier to cite or link to this item: http://hdl.handle.net/10045/107951
Información del item - Informació de l'item - Item information
Title: A soft computing approach to violence detection in social media for smart cities
Authors: Pujol, Francisco A. | Mora, Higinio | Pertegal-Felices, María Luisa
Research Group/s: UniCAD: Grupo de investigación en CAD/CAM/CAE de la Universidad de Alicante | Arquitecturas Inteligentes Aplicadas (AIA) | Habilidades, Competencias e Instrucción | Investigación en Inteligencias, Competencia Social y Educación (SOCEDU)
Center, Department or Service: Universidad de Alicante. Departamento de Tecnología Informática y Computación | Universidad de Alicante. Departamento de Psicología Evolutiva y Didáctica
Keywords: Violence detection | Smart cities | Social media | Radon transform | Optical acceleration
Knowledge Area: Arquitectura y Tecnología de Computadores | Didáctica y Organización Escolar
Issue Date: Aug-2020
Publisher: Springer Nature
Citation: Soft Computing. 2020, 24: 11007-11017. doi:10.1007/s00500-019-04310-x
Abstract: In recent years, social media has become an everyday tool for the distribution of videos in which signs of violence appear in different ways. Citizens of smart cities are demanding increasing efforts to authorities in order to maintain public safety, as well as to be efficient in an emergency response. The complexity of monitoring automatically the enormous amount of information generated through social networks results in the need for the development of systems that allow for the automatic detection of violent content in videos. This fact is becoming increasingly important in order to guarantee security for the citizens in any smart city. As a result, this work proposes the development of a system for detecting violence in videos by combining different descriptors that calculate the acceleration produced between two frames of a video. To do this, different techniques, such as the Radon transform or optical flow, are used. The trained system then performs the classification using support vector Machines. The results are promising, with accuracy rates between 85 and 97%, depending on the complexity of the databases used, which demonstrates the validity of our proposal.
Sponsor: This work has been partially supported by the Spanish Research Agency (AEI) and the European Regional Development Fund (FEDER) under project CloudDriver4Industry TIN2017-89266-R.
URI: http://hdl.handle.net/10045/107951
ISSN: 1432-7643 (Print) | 1433-7479 (Online)
DOI: 10.1007/s00500-019-04310-x
Language: eng
Type: info:eu-repo/semantics/article
Rights: © Springer-Verlag GmbH Germany, part of Springer Nature 2019
Peer Review: si
Publisher version: https://doi.org/10.1007/s00500-019-04310-x
Appears in Collections:INV - UNICAD - Artículos de Revistas
INV - AIA - Artículos de Revistas
INV - Habilidades, Competencias e Instrucción - Artículos de Revistas
INV - SOCEDU - Artículos de Revistas

Files in This Item:
Files in This Item:
File Description SizeFormat 
ThumbnailPujol_etal_2020_SoftComput_final.pdfVersión final (acceso restringido)1,57 MBAdobe PDFOpen    Request a copy

Items in RUA are protected by copyright, with all rights reserved, unless otherwise indicated.