Personalised video content summary and composition from ultra-large-volume video data driving by deep learning

  • Personalised video content summary and composition from ultra-large-volume video data driving by deep learning

School of Electronics, Electrical Engineering and Computer Science
& ECIT Global Research Institute

Proposed Project Title:
Personalised video content summary and composition from ultra-large-volume video data driving by deep learning

Principal Supervisor: Dr Yang Hua                   Second Supervisor:  Prof. Neil Robertson

Project Description:

Video has become one of the most popular visual media for entertainment, surveillance and communication in recent years. On the one hand, the volume of video data is exploding. For example, 300 hours of new videos are uploaded to Youtube every minute, and there are more than 4 million CCTVs in the UK alone, which produce an enormous amount of video data every day. On the other hand, end users would like to watch and retrieve specific content based on their own interests precisely and quickly. For example, a football fan will enjoy him-/herself for watching customised game highlights; A police officer can find specific suspicious events (such as car accident or robbery) from tons of surveillance videos.

To this end, the project targets on personalised video content summary and composition in terms of ultra-large-volume video data. By means of the advances in deep learning, the main components of this project are

  • Automatic video content summary and indexing based on specific criteria, e.g., length or relevance.
  • General scene and key person action understanding in video data.
  • Personalized video content composition according to user inputs or historical data.

Resource

As an industrial partner, BBC NI has engaged with the local Universities to select several innovative projects that they can collaborate with. The Personalised Video Content project proposal is of particular interest and if successful BBC NI will authorise the PhD student to access their video archive and support essential computation resource. The student may also have the opportunity to work within the world-class research team in BBC NI.



Contact details

Supervisor Name: Yang Hua                                                      Tel: +44 (0)28 9097 1816
QUB Address: Room 02-20, ECIT, Queens Road, BT3 9DT            Email: Y.Hua@qub.ac.uk