skip to main content
Idioma:

Hair detection, segmentation, and hairstyle classification in the wild

Muhammad, Umar Riaz ; Svanera, Michele ; Leonardi, Riccardo ; Benini, Sergio

Image and vision computing, 2018-03, Vol.71, p.25-37 [Periódico revisado por pares]

Elsevier B.V

Texto completo disponível

Citações Citado por
  • Título:
    Hair detection, segmentation, and hairstyle classification in the wild
  • Autor: Muhammad, Umar Riaz ; Svanera, Michele ; Leonardi, Riccardo ; Benini, Sergio
  • Assuntos: Hair database ; Hair detection ; Hair segmentation ; Hairstyle classification ; Texture analysis
  • É parte de: Image and vision computing, 2018-03, Vol.71, p.25-37
  • Descrição: Hair highly characterises human appearance. Hair detection in images is useful for many applications, such as face and gender recognition, video surveillance, and hair modelling. We tackle the problem of hair analysis (detection, segmentation, and hairstyle classification) from unconstrained view by relying only on textures, without a-priori information on head shape and location, nor using body-part classifiers. We first build a hair probability map by classifying overlapping patches described by features extracted from a CNN, using Random Forest. Then modelling hair (resp. non-hair) from high (resp. low) probability regions, we segment at pixel level uncertain areas by using LTP features and SVM. For the experiments we extend Figaro, an image database for hair detection to Figaro1k, a new version with more than 1000 manually annotated images. Achieved segmentation accuracy (around 90%) is superior to known state-of-the-art. Images are eventually classified into hairstyle classes: straight, wavy, curly, kinky, braids, dreadlocks, and short. •A workflow performing complete hair analysis (detection, segmentation and hairstyle classification) from unconstrained view is proposed, using only texture information without employing any body-part classifiers.•Figaro1k, a dataset of more than 1,000 unconstrained view images divided in seven hairstyle categories, is made publicly available along with the manually annotated hair masks.•Achieved segmentation accuracy (beyond 90%) is above performance reached by known state-of-the-art algorithms.•Automatic hairstyle recognition is performed for the first time by means of a multi-class texture classification step on the segmented hair region, obtained after the hair detection and segmentation phase.
  • Editor: Elsevier B.V
  • Idioma: Inglês
Voltar para lista de resultados

  • Realização: ABCD-USP USP   Logos de Redes Sociais: Freepik

Buscando em bases de dados remotas. Favor aguardar.