In this paper we propose the two-stage approach of organizing information invideo surveillance systems. At first, the faces are detected in each frame anda video stream is split into sequences of frames with face region of oneperson. Secondly, these sequences (tracks) that contain identical faces aregrouped using face verification algorithms and hierarchical agglomerativeclustering. Gender and age are estimated for each cluster (person) in order tofacilitate the usage of the organized video collection. The particularattention is focused on the aggregation of features extracted from each framewith the deep convolutional neural networks. The experimental results of theproposed approach using YTF and IJB-A datasets demonstrated that the mostaccurate and fast solution is achieved for matching of normalized average offeature vectors of all frames in a track.
展开▼
