Technology today has advanced to such a degree that we now have the unprecedented ability to record and store large volumes of data quite easily. However, interpreting this data in a high level and human-understandable manner requires a more erudite approach.

It is essential to help bridge the semantic gap between the raw data and the information required by users. We facilitate this by developing automated procedures to categorise and manage data sets, to find patterns and anomalies inherent in the data, and to assign contextual meaning and structure to the data.

The key research issues include:

• Scalability and Robust Systems for Behaviour Recognition. Our aim is to develop scalable systems in both spatial and temporal dimensions to understand and recognise human activity in a broad range of situations. The focus is on building hierarchical probabilistic models for learning and recognition. These systems should be able to recognise human behaviours and actions, understand and have an awareness of what is occurring in a particular scene, coordinate and recognise events related to the tracking of both individuals and groups and identify whole patterns based on partial patterns initially recognised by individual sensors.
• Bridging the Semantic Gap in Content Management. We focus on issues relating to the extraction of higher level semantics in video and film with a view to providing frameworks for both content acquisition and management.
  

For more details on current projects, please visit IMPCA's research page.