A targeted person searching scheme in digital videos based on face quality assessment and recognition

This research presents a targeted person searching scheme in digital videos. It is assumed that a user is provided with an exemplar video that contains a targeted person and long investigated videos from which the scenes related the targeted person will be extracted. First, the exemplar video is processed to generate multiple representative face images, which are shown on an interface for the user to manually select the images of a targeted person. After choosing the images characterizing the targeted person, the scheme will apply face tracking and face quality assessment for training a model, which can then help to search that person in other long investigated videos. The proposed scheme can facilitate such applications as evidence collection from surveillance videos and retrieval of actors from TV dramas or movies.