Gaze-based image retrieval system using dual eye-trackers

In this paper we present a novel gaze-based image retrieval application. The application is designed to be run on a dual monitor setup with a separate eye tracking device dedicated to each monitor. A source image is displayed on one monitor and the retrieved images are displayed on the second monitor. The system is completely gaze controlled. The user selects one or more objects or regions in the source image by fixating on them. The system then retrieves images containing similar objects from an image database. These are displayed in a grid on the second monitor. The user can then fixate on one of these images to select it as the new source image and the process can be repeated until a satisfactory image is found.