“…Very recently, open-world re-id [46], [47], [48] has been introduced, where persons in each camera may be only partially overlapping and the number of cameras, spatial size of the environment, and number of people may be unknown and at a significantly larger scale. Recall that the goal of this paper is to identify the persons given aligned images, which are the cases in most person re-identification benchmark datasets, while open-world re-id this is more a system level concept that must deal with issues such as person detection, tracking, re-id, data association, etc.…”