A Stochastic Grammar for Simultaneous Human Re-Identification and Tracking

Identifying humans across camera views is a long standing challenge especially while dealing with complex scenarios with frequent occlusions, significant lighting changes and other difficulties. Under such conditions, most existing appearance and geometric cues are not reliable enough to identify humans across camera views. To address these challenges, in this paper we propose to jointly re-identify and track humans of interest and introduce a stochastic attribute grammar model. The key idea of our method is to leverage complementary and discriminative human attributes for enhacing the above joint parsing task. In particular, we use a set of grammar rules to decompose a graph node (e.g. tracklet) into a set of children nodes (e.g. 3D human boxes), and augment each node with a set of attributes, including geometry (e.g., moving speed, direction), accessories (e.g., bags), and/or activities (e.g., walking, running). These attributes serve as valuable cues, in addition to appearance features (e.g., colors), in determining the associations of human detection boxes across cameras. The attributes of a parent node are inherited by its children nodes, resulting in consistency constraints over the feasible parse graph. Thus, we cast cross-view human re-identification and tracking as finding the most discriminative parse graph for each subject in videos. We develop a learning method to train this attribute grammar model from weakly supervised training data. We develop an alternative inference method to employ both topdown and bottom-up computations to search the optimal solution. We also explicitly reason the occlusion status of each entity in order to deal with significant changes of camera viewpoints. We evaluate the proposed method over public video benchmarks and demonstrate with extensive experiments that our method clearly outperforms state-of-theart tracking methods.