Distributed optimization for global data association in non-overlapping camera networks

Distributed data association in smart camera network via dual decomposition

“…(7), and the transition probability between neighboring cameras are set according to prior knowledge. Note here we use a weak but more realistic uniform model (8) instead of Gaussian model in previous works [1,12,32,43]. We set min , , 0.25…”

Multi-Object Tracking with Data Association by a Similarity Identification Model

“…In general, formulations with polynomial-time solutions consider evidence along paths of time-consecutive edges [25,41,109,166,167,265,329,372,374] and some build on bipartite matching [35,38,49,79,187,298,346]. Methods that use all the pairwise terms, not only those which are time-consecutive, are significantly more accurate but NP-hard [46,61,80,82,186,277,295,[316][317][318].…”

“…[6,255,260,348,358]. Many objects trackers, like [37] rely on extracting appearance information from the object pixels using hand-crafted features, where including appearance information via colour histogram [38,47,49,50,80,192,317,329,367] and texture descriptors [38,49,79,187,373,374] are the most popular representation for appearance modelling in MOT.…”

“…The first approaches to address MTMCT were inspired by the idea of connecting the tracking of the individuals from one sequence to the other by means of analysing the relationship between the cameras from which they were captured. Spatial relations between cameras have been either explicitly mapped in 3D [47,373], learned by tracking known identities [40,167,329], or obtained by comparing entry/exit rates across pairs of cameras [38,187]. Moreover, preprocessing methods have been used to fuse data from partially overlapping views [374], while some systems rely on completely overlapping and unobstructed views [12,25,32,173].…”

“…However, when the views are not overlapped, some works explicitly model people entry and exit points on the ground [38,47,187,228] or image plane [119,329]. Furthermore, travel time have been also modelled, either parametrically [329,373] or not [47,167,187]. For instance, Ristani et al [278] use time constraints to rule out unlikely inter-camera associations.…”