Abstract: Visual surveillance in wide areas (e.g. airports) relies on cameras that observe non-overlapping scenes. Multi-person tracking requires re-identification of people, when they leave one field of view and later enter another. For this we use appearance cues. Under the assumption that all observations of a single person are Gaussian distributed, the observation model in our approach consists of a Mixture of Gaussians: one component for each person. In this paper we propose a distributed approach for learning this MoG, where every camera learns from its own observations and the communication with other cameras. We propose a modified version of the recently developed Newscast EM algorithm for this. We test our alogithm on artificial generated data and on a collection of real-world observations gathered by a system of cameras in an office building.