LookSemantic Multimedia Analysis, Retrieval, and Transcoding (SMART)

Content-based image retrieval research has long realized the necessity of dealing with the Semantic Gap. We are interested in finding interesting patterns and semantics in image and video data, and developing and applying new machine learning techniques for problem-solving. Investigators of the project include Dr. Jeremiah D. Deng, Dr. Brendon Woodford, Prof. Martin Purvis, A/Prof. Stephen Cranefield.

Ongoing work

Signal analysis and sound classification

  • J. D. Deng, C. Simmermacher and S. Cranefield. A study on feature analysis for musical instrument classification. IEEE Trans. System, Man and Cybernetics - Part B, 38:429-438, 2008.
  • Other directions: mood recognition, melody retrieval

Sports Video Analysis

  • NEW: M. Shah, J.D. Deng, B.J. Woodford, Enhanced Codebook model for real-time background subtraction, Proc. of ICONIP'2011, LNCS vol.7064, 449-458, 2011.
  • M. Shah, J. D. Deng, B. J. Woodford, Localized Adaptive Learning of Mixture of Gaussians Models for Background Extraction, Proc. IVCNZ'10.
  • Other ongoing work: player detection and classification, event recognition

Semantic image analysis and novelty detection

  • SP Yong, J. D. Deng, M. K. Purvis, Novelty detection in wildlife scenes through semantic context modeling, accepted by Pattern Recognition, 2012.
  • SP Yong, J. D. Deng, M. K. Purvis, Modeling semantic context for novelty detection in wildlife scenes, ICME'10, pp.1254-1259, 2010.
  • Other directions: image/video near-duplicate detection

Video summarization / keyframe extraction

  • SP Yong, J. D. Deng, M. K. Purvis, Wildlife video key-frame extraction based on novelty detection in semantic context, Multimedia Tools and Applications, Springer, 2011, to appear.
  • D. Deng, Content-based image collection summarization and comparison using self-organizing maps. Pattern Recognition, 40:718-727, 2007.

Image/scene classification

  • J. D. Deng, Improving feature extraction for automatic medical image categorization, Proc. IVCNZ'09, Wellington, November 2009, pp.379-384.
  • J. D. Deng, R. Brinkworth, D. O'Carroll, Assessing the naturalness of scenes: an approach using statistics of local features, Image and Vision Computing New Zealand, 2008. IVCNZ 2008. 23rd International Conference, Christchurch, November 2008, 6 pages.

Coming up

Are you looking for a Honours/Masters topic? Consider:
  • Image and video super-resolution: how do we reuse our old SD video collections and show them on HD devices?
  • Semantics-assisted video transcoding and streaming: MPEG provides the platform but how do we take advantage of object recognition for better streaming?
  • EEG and music - Mozart or rap, does it make a difference?
Last modified: 8/2011.