Publicação

A machine learning approach to detect violent behaviour from video

Ver documento

Detalhes bibliográficos
Resumo:The automatic classification of violent actions performed by two or more persons is an important task for both societal and scientific purposes. In this paper, we propose a machine learning approach, based a Support Vector Machine (SVM), to detect if a human action, captured on a video, is or not violent. Using a pose estimation algorithm, we focus mostly on feature engineering, to generate the SVM inputs. In particular, we hand-engineered a set of input features based on keypoints (angles, velocity and contact detection) and used them, under distinct combinations, to study their effect on violent behavior recognition from video. Overall, an excellent classification was achieved by the best performing SVM model, which used keypoints, angles and contact features computed over a 60 frame image input range.
Autores principais:Nova, David
Outros Autores:Ferreira, André Leite; Cortez, Paulo
Assunto:Action recognition Machine learning Pose estimation Support Vector Machine Video analysis
Ano:2019
País:Portugal
Tipo de documento:comunicação em conferência
Tipo de acesso:acesso aberto
Instituição associada:Universidade do Minho
Idioma:inglês
Origem:RepositóriUM - Universidade do Minho
_version_ 1866876243668893696
author Nova, David
author2 Ferreira, André Leite
Cortez, Paulo
author2_role author
author
author_facet Nova, David
Ferreira, André Leite
Cortez, Paulo
author_role author
contributor_name_str_mv Universidade do Minho
country_str PT
creators_json_txt [{\"Person.name\":\"Nova, David\"},{\"Person.name\":\"Ferreira, André Leite\"},{\"Person.name\":\"Cortez, Paulo\"}]
datacite.contributors.contributor.contributorName.fl_str_mv Universidade do Minho
datacite.creators.creator.creatorName.fl_str_mv Nova, David
Ferreira, André Leite
Cortez, Paulo
datacite.date.Accepted.fl_str_mv 2019-01-01T00:00:00Z
datacite.date.available.fl_str_mv 2019-12-20T14:47:15Z
datacite.date.embargoed.fl_str_mv 2019-12-20T14:47:15Z
datacite.rights.fl_str_mv http://purl.org/coar/access_right/c_abf2
datacite.subjects.subject.fl_str_mv Action recognition
Machine learning
Pose estimation
Support Vector Machine
Video analysis
datacite.titles.title.fl_str_mv A machine learning approach to detect violent behaviour from video
dc.contributor.none.fl_str_mv Universidade do Minho
dc.creator.none.fl_str_mv Nova, David
Ferreira, André Leite
Cortez, Paulo
dc.date.Accepted.fl_str_mv 2019-01-01T00:00:00Z
dc.date.available.fl_str_mv 2019-12-20T14:47:15Z
dc.date.embargoed.fl_str_mv 2019-12-20T14:47:15Z
dc.format.none.fl_str_mv application/pdf
dc.identifier.none.fl_str_mv https://hdl.handle.net/1822/62744
dc.language.none.fl_str_mv eng
dc.publisher.none.fl_str_mv Springer Verlag
dc.rights.none.fl_str_mv http://purl.org/coar/access_right/c_abf2
dc.subject.none.fl_str_mv Action recognition
Machine learning
Pose estimation
Support Vector Machine
Video analysis
dc.title.fl_str_mv A machine learning approach to detect violent behaviour from video
dc.type.none.fl_str_mv http://purl.org/coar/resource_type/c_5794
description The automatic classification of violent actions performed by two or more persons is an important task for both societal and scientific purposes. In this paper, we propose a machine learning approach, based a Support Vector Machine (SVM), to detect if a human action, captured on a video, is or not violent. Using a pose estimation algorithm, we focus mostly on feature engineering, to generate the SVM inputs. In particular, we hand-engineered a set of input features based on keypoints (angles, velocity and contact detection) and used them, under distinct combinations, to study their effect on violent behavior recognition from video. Overall, an excellent classification was achieved by the best performing SVM model, which used keypoints, angles and contact features computed over a 60 frame image input range.
dirty 0
eu_rights_str_mv openAccess
format conferencePaper
fulltext.url.fl_str_mv https://prod-dspace.uminho.pt/bitstreams/cc86e7fc-2bea-4202-b13a-8ec1c207d054/download
id rum_2e6ccb062cc33ec7adc81636be5a2fef
identifier.url.fl_str_mv https://hdl.handle.net/1822/62744
instacron_str repositorium
institution Universidade do Minho
instname_str Universidade do Minho
language eng
network_acronym_str rum
network_name_str RepositóriUM - Universidade do Minho
oai_identifier_str oai:repositorium.uminho.pt:1822/62744
organization_str_mv urn:organizationAcronym:repositorium
person_str_mv Nova, David
Ferreira, André Leite
Cortez, Paulo
publishDate 2019
publisher.none.fl_str_mv Springer Verlag
reponame_str RepositóriUM - Universidade do Minho
repository_id_str urn:repositoryAcronym:rum
service_str_mv urn:repositoryAcronym:rum
spelling engSpringer VerlagporThe automatic classification of violent actions performed by two or more persons is an important task for both societal and scientific purposes. In this paper, we propose a machine learning approach, based a Support Vector Machine (SVM), to detect if a human action, captured on a video, is or not violent. Using a pose estimation algorithm, we focus mostly on feature engineering, to generate the SVM inputs. In particular, we hand-engineered a set of input features based on keypoints (angles, velocity and contact detection) and used them, under distinct combinations, to study their effect on violent behavior recognition from video. Overall, an excellent classification was achieved by the best performing SVM model, which used keypoints, angles and contact features computed over a 60 frame image input range.application/pdfporA machine learning approach to detect violent behaviour from videoNova, DavidFerreira, André LeiteCortez, PauloHostingInstitutionOrganizationalUniversidade do Minhoe-mailmailto:repositorium@usdb.uminho.ptrepositorium@usdb.uminho.ptISBNIsPartOf9783030164461ISSNIsPartOf1867-8211DOIIsPartOf10.1007/978-3-030-16447-8_92019-12-20T14:47:15Z20192019-12-20T14:34:27Z2019-01-01T00:00:00ZHandlehttps://hdl.handle.net/1822/62744http://purl.org/coar/access_right/c_abf2open accessAction recognitionMachine learningPose estimationSupport Vector MachineVideo analysis2106683 bytesother research producthttp://purl.org/coar/resource_type/c_5794conference paperhttp://purl.org/coar/access_right/c_abf2application/pdffulltexthttps://prod-dspace.uminho.pt/bitstreams/cc86e7fc-2bea-4202-b13a-8ec1c207d054/download
spellingShingle A machine learning approach to detect violent behaviour from video
Nova, David
Action recognition
Machine learning
Pose estimation
Support Vector Machine
Video analysis
status SINGLETON
subject.fl_str_mv Action recognition
Machine learning
Pose estimation
Support Vector Machine
Video analysis
title A machine learning approach to detect violent behaviour from video
title_full A machine learning approach to detect violent behaviour from video
title_fullStr A machine learning approach to detect violent behaviour from video
title_full_unstemmed A machine learning approach to detect violent behaviour from video
title_short A machine learning approach to detect violent behaviour from video
title_sort A machine learning approach to detect violent behaviour from video
topic Action recognition
Machine learning
Pose estimation
Support Vector Machine
Video analysis
topic_facet Action recognition
Machine learning
Pose estimation
Support Vector Machine
Video analysis
url https://hdl.handle.net/1822/62744
visible 1