A Novel System Solution for Crowd Supervision

Abstract

In this paper, we introduce a novel system solution for crowd supervision. Different from other solutions based on algorithms, in this paper we propose to build a system solution for congestion detection and early warning, based on the real monitoring scene and the status of existing software and hardware for video big data. The software and hardware of the scheme are divided into three layers, and the specific solution is divided into seven steps. The scheme has been applied to some cities in China, such as Shenzhen and Nanjing, and has achieved good performance on large-scale activities. More importantly, this solution is mainly based on unsupervised algorithms, and hence is readily scalable to realworld large-scale scenarios, and is more flexible to apply to different places. In other words, our solution does not need to exhaustively collect a large number of cross-view pairwise labels for each camera pair as required by most existing solutions.