Multiple Objects Association System for the Smart City
Aleksey Popov, Stanislav V. Ibragimov, Sergei A. Malyshev, Rufina A. Abdurakhmanova · 2021
The increase of population in large cities require safety systems improvement for urban environment. The traditional Smart City security systems are highly centralized to process a large number of video streams in the massive-parallel data centers. As a result, this requires an expensive network infrastructure and makes it available only for large metropolitan areas. In this paper, we propose a distributed system for analyzing the luggage and people relations in crowded places. The system processes video on the camera's nodes and detects dependent tracks for objects moving. Then we form a combined scene in the central node to accurately represent the association of people and luggage. In this research stage, we present the basics concepts of multiple object re-identification and association measurement, as well as knowledge graph constructing. We describe the methodology of data obtaining and show experimental results of visual object recognition on camera nodes.