Construction of co-occurrence relationship between pedestrians based on smart face capture cameras’ data
Shiqi Qu, Yuning Liu, Bingbing Fan · 2021
With the continuous advancement of face recognition technology and the large-scale and intensive deployment of smart face capture cameras, this paper conducts in-depth mining of its data, explores the co-occurrence relationship between pedestrians, and provides clues when there is a need to obtain someone’s real social relationship. After experimental comparison and demonstration, this paper uses Chinese Whispers clustering algorithm to identify pedestrians nodes, and then use Faiss to accelerate its construction of adjacent edges, thereby speeding up the initialization step of the graph, and solving the low problem of clustering efficiency. On this basis, co-occurrence frequency and the concept of confidence in Apriori algorithm are used to mine the co-occurrence relationship between pedestrians, and build a pedestrian co-occurrence relationship graph.