作者: David López-Fernández , Francisco José Madrid-Cuevas , Ángel Carmona-Poyato , Manuel Jesús Marín-Jiménez , Rafael Muñoz-Salinas
DOI: 10.1007/978-3-319-13323-2_3
关键词:
摘要: In this paper, we introduce a new multi-view dataset for gait recognition. The was recorded in an indoor scenario, using six convergent cameras setup to produce videos, where each video depicts walking human. Each sequence contains at least 3 complete cycles. videos of 20 persons with large variety body size, who walk along straight and curved paths. have been processed foreground silhouettes. To validate our dataset, extended some appearance-based 2D recognition methods work 3D data, obtaining very encouraging results. as well camera calibration information, is freely available research purposes.