Abstract:
In this project we, propose a novel eye-gaze tracking interface that can work with a normal HD web camera. The developed framework can be used to create eye movement datasets, running behavioral experiments with images, and gaze tracking. This facilitates behavioral experiments and research in visual cognition. The developed framework also enables the user to select images, display them for specific period and record corresponding (fixations and saccades). There is also a built in calibration tool for estimating the gaze error. The developed framework uses HAAR cascade classifier to detect face and then localize eye in the region of interest. The process is robust to different light conditions with throughput of 25 to 30 FPS. The framework also enables the user to annotate the experimental data and/or stimulus with subject specific information (such as name, age, gender, profession and interests). The experiment data for each run of experiment is saved in a binary/text file for further processing.
