Overview
Topic status: We're looking for students to study this topic.
Project Summary
This project will investigate the use of soft biometrics and basic scene recognition to redetect persons across video cuts or scene changes. In video analysis and indexing applications, persons of interest often come in and out of view multiple times due to the editing and composition of the video. It is desirable to treat multiple appearances of a person as one individual, rather than different individuals.
Biometrics are measurements of the personally identifiable physical characteristics of persons. Traditional biometric recognition such as face recognition can fail for many reasons, such as large head pose change, expression change or lighting change. Using multiple biometrics can help alleviate the problem, but capturing biometrics from video is difficult. Soft biometrics are traits that are not unique, but can be used to identify a person in a constrained context. Examples include hair, skin and clothing colour and height. This project will employ these soft biometrics to aid redetection of persons where face recognition fails. Basic scene recognition methods will also be used to identify reoccurring scenes containing the same persons.
Expected outcomes, applications and/or benefits
A basic understanding of multiple biometrics, soft biometrics and scene recognition.
Improved ability to associate observations of a person across multiple video cuts with the same identity.
Required student skills/experience
Strong C++ programming skills.
- Study level
- Vacation research experience scholarship
- Supervisors
- QUT
- Organisational unit
Science and Engineering Faculty
- Research area
- Keywords
- redetection, biometrics, recognition
- Contact
- Contact the supervisor for more information
Professor Sridha Sridharan
Associate Professor Clinton Fookes
Dr Simon Denman
