%0 Conference Proceedings
%4 sid.inpe.br/sibgrapi/2016/09.09.13.16
%2 sid.inpe.br/sibgrapi/2016/09.09.13.16.25
%T An approach for Brazilian Sign Language (BSL) recognition based on facial expression and k-NN classifier
%D 2016
%A Rezende, Tamires Martins,
%A Castro, Cristiano Leite de,
%A Almeida, Sílvia Grasiella M.,
%@affiliation The Electrical Engineering Graduate Program - Federal University of Minas Gerais - Brazil
%@affiliation The Electrical Engineering Graduate Program - Federal University of Minas Gerais - Brazil
%@affiliation Department of Industrial Automation - Federal Institute of Minas Gerais - Ouro Preto, Brazil
%E Aliaga, Daniel G.,
%E Davis, Larry S.,
%E Farias, Ricardo C.,
%E Fernandes, Leandro A. F.,
%E Gibson, Stuart J.,
%E Giraldi, Gilson A.,
%E Gois, João Paulo,
%E Maciel, Anderson,
%E Menotti, David,
%E Miranda, Paulo A. V.,
%E Musse, Soraia,
%E Namikawa, Laercio,
%E Pamplona, Mauricio,
%E Papa, João Paulo,
%E Santos, Jefersson dos,
%E Schwartz, William Robson,
%E Thomaz, Carlos E.,
%B Conference on Graphics, Patterns and Images, 29 (SIBGRAPI)
%C São José dos Campos, SP, Brazil
%8 4-7 Oct. 2016
%I Sociedade Brasileira de Computação
%J Porto Alegre
%S Proceedings
%K RGB-D sensor, Brazilian Sign Language, k-NN, Facial expression.
%X The automatic recognition of facial expressions is a complex problem that requires the application of Computational Intelligence techniques such as pattern recognition. As shown in this work, this technique may be used to detect changes in physiognomy, thus making it possible to differentiate between signs in BSL (Brazilian Sign Language or LIBRAS in Portuguese). The methodology for automatic recognition in this study involved evaluating the facial expressions for 10 signs (to calm down, to accuse, to annihilate, to love, to gain weight, happiness, slim, lucky, surprise, and angry). Each sign was captured 10 times by an RGB-D sensor. The proposed recognition model was achieved through four steps: (i) detection and clipping of the region of interest (face), (ii) summarization of the video using the concept of maximized diversity, (iii) creation of the feature vector and (iv) sign classification via k-NN (k-Nearest Neighbors). An average accuracy of over 80\% was achieved, revealing the potential of the proposed model.
%@language en
%3 6.pdf