Locating People in Video Surveillance from Semantic Descriptions
Michael Halstead · Queensland University of Technology · 2017
This PhD research developed new approaches to locate individuals in surveillance footage from a semantically describable target description. Investigations in this programme focused on the use of soft biometrics such as clothing colours and patterns, and height as traits to locate a target subject in video. This facilitated the introduction of techniques able to reduce the impact on human operators in surveillance settings.