Facial expressions: emotions recognition based on selected landmarks

Detalhes bibliográficos
Autor(a) principal: Lopes, Daniela Soares de Pina
Data de Publicação: 2022
Tipo de documento: Dissertação
Idioma: eng
Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
Texto Completo: http://hdl.handle.net/10773/36998
Resumo: Automatic emotion recognition based on facial expressions is an active and challenging research field, owning to its academic and commercial potential. Most of the already existing facial expression recognition systems use a holistic approach to withdraw emotional characteristics from facial images, i.e., they adopt methods that make use of all facial attributes in order to recognize emotions. This project aims to identify the regions of the face that characterize each emotion within a set of basic emotions. For this, research was conducted to study the facial movements that occur from the neutral expression to the apex of a given emotion, resorting to the analysis of the evolution of the euclidean distances between facial landmarks throughout the frames of a video. From this analysis, there were selected sets of facial landmarks essential to identify each facial expression. The model used to perform the facial expressions classification was the Support Vector Machine (SVM). During this project, it was possible to verify that the number of facial landmarks can be reduced to a minimum value, promoting better computational efficiency, the usage of simpler algorithms, and also, in case it is necessary to store information, preserve people’s privacy since it diminishes the probability of re-identifying the intervening subjects with only the selected features.
id RCAP_95f1bc67d7868cdb9f0796676762a025
oai_identifier_str oai:ria.ua.pt:10773/36998
network_acronym_str RCAP
network_name_str Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
repository_id_str 7160
spelling Facial expressions: emotions recognition based on selected landmarksAffective computingFacial expressions recognitionFacial landmarksImage processingMachine learningAutomatic emotion recognition based on facial expressions is an active and challenging research field, owning to its academic and commercial potential. Most of the already existing facial expression recognition systems use a holistic approach to withdraw emotional characteristics from facial images, i.e., they adopt methods that make use of all facial attributes in order to recognize emotions. This project aims to identify the regions of the face that characterize each emotion within a set of basic emotions. For this, research was conducted to study the facial movements that occur from the neutral expression to the apex of a given emotion, resorting to the analysis of the evolution of the euclidean distances between facial landmarks throughout the frames of a video. From this analysis, there were selected sets of facial landmarks essential to identify each facial expression. The model used to perform the facial expressions classification was the Support Vector Machine (SVM). During this project, it was possible to verify that the number of facial landmarks can be reduced to a minimum value, promoting better computational efficiency, the usage of simpler algorithms, and also, in case it is necessary to store information, preserve people’s privacy since it diminishes the probability of re-identifying the intervening subjects with only the selected features.O reconhecimento automático de emoções baseado em expressões faciais é uma área de pesquisa ativa e desafiante, fazendo jus ao seu potencial académico e comercial. A maioria dos sistemas de reconhecimento de expressões faciais já existentes utilizam uma abordagem holística para retirar características emocionais da face, ou seja, adoptam métodos que fazem uso de todos os atributos faciais para reconhecer emoções. Este projeto visa identificar as regiões do rosto que caracterizam cada emoção dentro de um conjunto de emoções básicas. Para tal, foi conduzida uma pesquisa para estudar os movimentos faciais que ocorrem desde a expressão neutra até ao ápice de uma determinada emoção, recorrendo à análise da evolução das distâncias euclidianas entre os pontos de referência na face ao longo dos frames de um vídeo. A partir desta análise, foram selecionados conjuntos de pontos de referência essenciais para identificar cada expressão facial. O modelo utilizado para realizar a classificação das expressões faciais foi o Support Vector Machine (SVM). No decorrer deste trabalho foi possível verificar que o número de pontos descritores da face pode ser reduzido a um valor mínimo, possibilitando assim uma maior eficiência computacional, o uso de algoritmos mais simples e ainda, no caso de ser necessária guardar informação, preservar a privacidade das pessoas, reduzindo a probabilidade de re-identificar os intervenientes com base nas features selecionadas.2023-04-13T14:52:12Z2022-12-15T00:00:00Z2022-12-15info:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/masterThesisapplication/pdfhttp://hdl.handle.net/10773/36998engLopes, Daniela Soares de Pinainfo:eu-repo/semantics/openAccessreponame:Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)instname:Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informaçãoinstacron:RCAAP2024-02-22T12:10:48Zoai:ria.ua.pt:10773/36998Portal AgregadorONGhttps://www.rcaap.pt/oai/openaireopendoar:71602024-03-20T03:07:26.433695Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) - Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informaçãofalse
dc.title.none.fl_str_mv Facial expressions: emotions recognition based on selected landmarks
title Facial expressions: emotions recognition based on selected landmarks
spellingShingle Facial expressions: emotions recognition based on selected landmarks
Lopes, Daniela Soares de Pina
Affective computing
Facial expressions recognition
Facial landmarks
Image processing
Machine learning
title_short Facial expressions: emotions recognition based on selected landmarks
title_full Facial expressions: emotions recognition based on selected landmarks
title_fullStr Facial expressions: emotions recognition based on selected landmarks
title_full_unstemmed Facial expressions: emotions recognition based on selected landmarks
title_sort Facial expressions: emotions recognition based on selected landmarks
author Lopes, Daniela Soares de Pina
author_facet Lopes, Daniela Soares de Pina
author_role author
dc.contributor.author.fl_str_mv Lopes, Daniela Soares de Pina
dc.subject.por.fl_str_mv Affective computing
Facial expressions recognition
Facial landmarks
Image processing
Machine learning
topic Affective computing
Facial expressions recognition
Facial landmarks
Image processing
Machine learning
description Automatic emotion recognition based on facial expressions is an active and challenging research field, owning to its academic and commercial potential. Most of the already existing facial expression recognition systems use a holistic approach to withdraw emotional characteristics from facial images, i.e., they adopt methods that make use of all facial attributes in order to recognize emotions. This project aims to identify the regions of the face that characterize each emotion within a set of basic emotions. For this, research was conducted to study the facial movements that occur from the neutral expression to the apex of a given emotion, resorting to the analysis of the evolution of the euclidean distances between facial landmarks throughout the frames of a video. From this analysis, there were selected sets of facial landmarks essential to identify each facial expression. The model used to perform the facial expressions classification was the Support Vector Machine (SVM). During this project, it was possible to verify that the number of facial landmarks can be reduced to a minimum value, promoting better computational efficiency, the usage of simpler algorithms, and also, in case it is necessary to store information, preserve people’s privacy since it diminishes the probability of re-identifying the intervening subjects with only the selected features.
publishDate 2022
dc.date.none.fl_str_mv 2022-12-15T00:00:00Z
2022-12-15
2023-04-13T14:52:12Z
dc.type.status.fl_str_mv info:eu-repo/semantics/publishedVersion
dc.type.driver.fl_str_mv info:eu-repo/semantics/masterThesis
format masterThesis
status_str publishedVersion
dc.identifier.uri.fl_str_mv http://hdl.handle.net/10773/36998
url http://hdl.handle.net/10773/36998
dc.language.iso.fl_str_mv eng
language eng
dc.rights.driver.fl_str_mv info:eu-repo/semantics/openAccess
eu_rights_str_mv openAccess
dc.format.none.fl_str_mv application/pdf
dc.source.none.fl_str_mv reponame:Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
instname:Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informação
instacron:RCAAP
instname_str Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informação
instacron_str RCAAP
institution RCAAP
reponame_str Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
collection Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
repository.name.fl_str_mv Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos) - Agência para a Sociedade do Conhecimento (UMIC) - FCT - Sociedade da Informação
repository.mail.fl_str_mv
_version_ 1799137729375633408