Facial Landmarks Detection on Faulty Datasets with Regression Trees and Principal Component Analysis Parametrization

  • Abner S. Nascimento UFC
  • Danilo A. Oliveira UFC
  • Maria Raquel L. de Couto UFC
  • Iális C. de Paula Júnior UFC

Resumo


Tracking landmarks points of the human face is an essential step for the construction of interfaces capable of taking advantage of the communicative potential of facial expressions. Many strategies based on parametric models and regression algorithms with boosting can be applied to this problem. This paper proposes a solution based on the combined use of principal component analysis and regression trees. The main purpose of the presented method is to reduce the sensibility of the system to the presence of missing labels when trained with faulty datasets, by the adoption of corrective heuristics. On such cases, the proposed model achieves performance similar to the reference results, obtained by training on fault free datasets.

Referências


Cao, X., Wei, Y., Wen, F., e Sun, J. (2012). Face alignment by explicit shape regression. Em 2012 IEEE Conference on Computer Vision and Pattern Recognition, páginas 2887–2894.

Cootes, T. F., Edwards, G. J., e Taylor, C. J. (2001). Active appearance models. IEEE Transactions on Pattern Analysis & Machine Intelligence, (6):681–685.

Cootes, T. F., Hill, A., Taylor, C. J., e Haslam, J. (1994). Use of active shape models for locating structures in medical images. Image and vision computing, 12(6):355–366.

Dalal, N. e Triggs, B. (2005). Histograms of oriented gradients for human detection. Em Computer Vision and Pattern Recognition, 2005. CVPR 2005. IEEE Computer Society Conference on, volume 1, páginas 886–893. IEEE.

Dollár, P., Welinder, P., e Perona, P. (2010). Cascaded pose regression. Em Computer Vision and Pattern Recognition (CVPR), 2010 IEEE Conference on, páginas 1078–1085. IEEE.

Frith, C. (2009). Role of facial expressions in social interactions. Philosophical transactions of the royal society of London B: Biological sciences, 364(1535):3453–3458.

Hill, T. e Lewicki, P. (2006). Statistics: Methods and Applications: A Comprehensive Reference for Science, Industry, and Data Mining.

Jaimes, A. e Sebe, N. (2007). Multimodal human–computer interaction: A survey. Computer vision and image understanding, 108(1-2):116–134.

Kazemi, V. e Sullivan, J. (2014). One millisecond face alignment with an ensemble of regression trees. Em Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, páginas 1867–1874.

Kendall, D. G. (1989). A survey of the statistical theory of shape. Statistical Science, 4(2):87–99.

Le, V., Brandt, J., Lin, Z., Bourdev, L., e Huang, T. S. (2012). Interactive facial feature localization. Em European Conference on Computer Vision, páginas 679–692. Springer.

Zhu, X. e Ramanan, D. (2012). Face detection, pose estimation, and landmark localization in the wild. Em Computer Vision and Pattern Recognition (CVPR), 2012 IEEE Conference on, páginas 2879–2886. IEEE.

Publicado
22/10/2018
NASCIMENTO, Abner S.; OLIVEIRA, Danilo A.; DE COUTO, Maria Raquel L.; DE PAULA JÚNIOR, Iális C.. Facial Landmarks Detection on Faulty Datasets with Regression Trees and Principal Component Analysis Parametrization. In: ENCONTRO NACIONAL DE INTELIGÊNCIA ARTIFICIAL E COMPUTACIONAL (ENIAC), 15. , 2018, São Paulo. Anais [...]. Porto Alegre: Sociedade Brasileira de Computação, 2018 . p. 343-352. ISSN 2763-9061. DOI: https://doi.org/10.5753/eniac.2018.4429.