Título: Improvements on head pose estimation algorithm for state classification
Autores: Silva, Renato Ramos da; Romero, Roseli A. Francelin
Resumo: An important aspect for a robot to interact with humans is to detect a person in the environment with which it is interacting and to estimate her/him head pose. These aspects made possible to a robot know if it will alone and execute some work or interact socially with a person doing a job or by fun. There are several works of head pose estimation in the literature, but we focus on work that used local histogram of oriented gradient and support vector regression to solve this problem. In this paper, we are proposing an adaptation of this method proposed using only histogram of oriented gradient in order to classify the head pose estimation and the presence of a person in the environment. Then, the new method has been evaluated in a offline and online way. The results showed that the proposed approach has great capability to classify head pose and identify when a person is present in the environment.
Código DOI: 10.21528/CBIC2013-030
Artigo em pdf: bricsccicbic2013_submission_30.pdf
Arquivo BibTex: bricsccicbic2013_submission_30.bib