Detecção de maquiagem facial por meio de CMYK e redes neurais

• Ano de defesa: 2018
• Autor(a) principal: Bertacchi, Marcello Guariento
• Orientador(a): Silveira, Ismar Frango
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Presbiteriana Mackenzie
• Palavras-chave em Português: detecção de maquiagem; visão computacional; CMYK; redes neurais; HSV
• Área do conhecimento CNPq: CNPQ::ENGENHARIAS
• Link de acesso: http://dspace.mackenzie.br/handle/10899/24475

Initially, facial feature recognition was only used intuitively, which means that one individual recognized another by certain characteristics relevant for their identification. Time passed, and with technological advancement, other methods were created for this purpose. However, the addition of artificial characteristics could have a negative influence in the process of facial recognition. Hence the choice of the cosmetic application field, with the purpose of exploring in more details both the effects in recognition as well as the process of detection of facial makeup. For this purpose, the color model CMYK was chosen due to its satisfactory performance in skin detection. The objective of this work is to emphasize the feasibility of applying the color model CMYK in Computational Vision procedures and Image Analysis, in comparisson to another model widely used, which is the HSV. For the makeup classification process, it was chosen a variant of Artificial Neural Networks known as Neural Network Convolutional, which is based on the visual cortex of cats. First, it was proved the negative influence of makeup in face recognition, through the LBP descriptor. In sequence, six neural networks were trained to detect makeup, achieving an accuracy of 97 percentage points on the eye region, 95 points percent on the face and 80 percentage points on the lips, in CMYK’s model, and 91 percentage points on the eye region, 92 points percent on the face and 73 percentage points on the lips, in HSV’s model. Consequently, CMYK was proven to be a color space that deserves attention in the fields of Makeup and Computer Vision.