Fellowship focus: Solving a fundamental problem in computer vision by developing a method for segmenting objects found in images in real-world scenarios.
At present, computer vision systems are able to segment predefined objects (for example they can be trained to recognise ‘dogs’ and ‘cats’) but often fail when confronted by new classes of object (such as ‘pandas’). Associate Professor Zhang will spend two months at Surrey with the aim of building a new multimodal framework that combines images, text and sound, and enables systems to adapt quickly to new or poorly-trained classes of object. The Fellowship will include the establishment of a research network with China University of Petroleum and a jointly-organised seminar on computer vision and AI.