Gan-based realistic face pose synthesis

• Ano de defesa: 2018
• Autor(a) principal: Souza, Douglas Matos de
• Orientador(a): Ruiz, Duncan Dubugras Alcoba
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: eng
• Instituição de defesa: Pontifícia Universidade Católica do Rio Grande do Sul
• Programa de Pós-Graduação: Programa de Pós-Graduação em Ciência da Computação
• Departamento: Escola Politécnica
• País: Brasil
• Palavras-chave em Português: Síntese de poses de faces; Aprendizado de máquina; Aprendizado profundo; Visão computacional; Redes geradoras adversárias
• Palavras-chave em Inglês: Face pose synthesis; Machine learning; Deep learning; Computer vision; Generative adversarial networks
• Área do conhecimento CNPq: CIENCIA DA COMPUTACAO::TEORIA DA COMPUTACAO
• Link de acesso: http://tede2.pucrs.br/tede2/handle/tede/9250

In computer vision, processing face images are accompanied by a series of complexities. Examples include variation of pose, light, face expression, and make up. Although all aspects are considered important, the one that impacts the most face-related computer vision systems is pose. In face recognition, for example, it has been long desired to have a method capable of bringing faces to the same pose, usually a frontal view, in order to ease recognition. Synthesizing different views of a face is a great challenge, mostly because in non-frontal face images there are loss of information when one side of the face occludes the other (also known as self-occlusion). Several methods to address face pose synthesis were proposed, but the results usually miss a realistic finish. In this work, we present novel methods that improve on the previous ones, showing higher synthesis quality.