Gaze Tracking -Shashank Shekhar
Aim ● To estimate a person's gaze using a webcam. ● Gaze tracking can be used in interactive applications as a means to take inputs.
Possible Approaches ● 3-D gaze tracking ● 2-D gaze tracking ● Estimates 3-D gaze ● Uses a 2-D mapping and find the point of function to map gaze intersection with from image to screen screen ● Requires near IR light ● Requires stereo or high resolution camera. webcam
2-D gaze tracking ● Uses pupil-glint vector and maps it to screen point by using following eqns – Xs = a + a1 * Vx + a2 * Vy + a3 * Vx *Vy – Ys = b + b1 * Vx + b2 * Vy + b3 * Vy *Vy –
Challenges ● Extraction of glint is fairly easy by using a strong light source. ● Extraction of pupil from iris as background is tricky and requires a fairly high resolution camera.
Work Done till Now ● Head and eye detection using Haar cascades ● The image after using canny edge detector clearly shows the glint but pupil extraction would require some work.
Questions ??
Recommend
More recommend
