Features Accumulation on a Multiple View Oriented Model for People Re-Identification
Jorge García, Chandra Kambhamettu, Alfredo Gardel, Ignácio Bravo, Lázaro, J. L. · Eurographics Workshop on 3D Object Retrieval · 2013
People re-identification process provides relevant information in order to understand the scene. In this paper, we present a multiple-view oriented model for performing people re-identification in a camera network. An appear- ance model for different perspectives is generated from people trajectories. Global and local features besides path orientation are extracted from each person's image, given a short-term tracking. Experimental results based on different sets of features demonstrate the effectiveness of our system.