Creating machines that ‘see’ and ‘understand’ the world around them
Making machines with the intelligence to ‘see’ and ‘understand’ the physical world is the focus of this research stream. Recognised as a world-leader, we investigate new technologies and robust systems for face recognition, facial expression understanding, image retrieval, biometrics and 2D and 3D object recognition. Recent research has also focussed on insect identification and interpretation, medical image analysis, scene understanding and modelling.
Furthermore, we develop systems exhibiting human speech and visual perception, investigating robust systems for speech recognition, speech coding, speech enhancement, speech synthesis, and speaker identification.
- Face detection and recognition
- Facial expression understanding
- Multi-modal fusion and recognition
- String matching
- Multimedia information retrieval
- Biomedical engineering
- Insect identification and interpretation
- Fire Ant screening and identification
- Human-Computer Communication with Speech and Image Modalities
Automatic Identification and Screening of Red Imported Fire Ants
With the support of the Queensland and Federal Governments, research is conducted for developing a prototype screening system to quickly identify red imported fire ants. These ants are not only extremely destructive, but also tiny and impossible to identify with the naked eye. Easily confused with common coastal brown ants and other native ants, their correct identification currently involves biosecurity officers travelling to areas of suspected ant outbreaks to collect samples and bring them back to the lab to identify them under a microscope. This is time consuming and expensive. We have developed a remote microscope will collect images to a laptop or PDA where the screening system will automatically perform species identification in situ with computer vision and pattern recognition technologies.
Directional Corner Points for Robust Face Retrieval
This research proposes a robust face retrieval approach using structural and spatial point correspondence in which the Directional Corner Points (DCP) are generated for efficient face coding and retrieval. Experimental results demonstrate that the proposed DCP approach provides a new way, which is both robust to scale and environmental changes, and efficient in computation, for retrieving human faces in single model databases.