Generalized Image Retrieval and Relation Discovery – a new Grant

Ondrej Chum received Grant

Generalized Image Retrieval and Relation Discovery
In the project, will be generalized the retrieval of related images beyond search of visually similar images. Team will study various definitions of relatedness, investigate algorithms for efficient search, and derive novel representations that support search and discovery of related images. Applications range from visual localization under significant change in view point and illumination (day and night), through the diversity retrieval in changing environments (changes in urban areas or evolution of a glacier),multi-modal search (sketch based image retrieval), to case retrieval in medicine (finding similar examination imaging of other patients to improve the diagnosis).

Responsible person: Petra Rosická