Effectual Training for Object Detection Using Eye Tracking Data Set

Eye tracking device is playing a major role in research area due to its feature to capture the eye movements of a human. Eye tracker gives the patterns of visual attention of the observer which is analyzed in terms of fixations and saccades. The proposed work is used eye tracker for training the system of automatic detection of no-entry sign board. The whole process is divided into three stages: in first stage, fixation point is collected from an eye-tracker device. In second stage, objects are tagged by k-means clustering algorithm applied on fixation points. In third stage, the system is trained using cascade training algorithm to detect the objects automatically. No-entry sign boards are considered as the target in the proposed work. Detection of objects is based on fixation points rather than analyzing the features of all the pixels of the images.