Researchers at the King Abdullah University of Science and Technology (KAUST) and the University College London have proposed a tool that generates image queries based on a sketch or description of objects in spatial relationships, which could make it easier to search large databases of images. Instead of only describing the individual objects in an image, the researchers want to develop a method for describing the relationships between objects that can be computed and searched efficiently, says KAUST researcher Peter Wonka. The researchers developed a query tool called relation-augmented image descriptor (RAID), which takes either a written description or sketch of objects in a specific spatial relationship and searches for matches in the image database based on geometric processing. "RAID allows us to search using a sentence such as 'person standing on snowboard' or to use a simple sketch of the desired composition of objects or an example image with the desired object composition," Wonka says. He notes the new method uses a novel description based on the spatial distribution of relationships over the entire object, which enables users to successfully discriminate between different complex relationships. Wonka says RAID provides a new way to describe images and could be applied to computer graphics, computer vision, and automated object classification.