SINC(I)   25518
INSTITUTO DE INVESTIGACION EN SEÑALES, SISTEMAS E INTELIGENCIA COMPUTACIONAL
Unidad Ejecutora - UE
congresos y reuniones científicas
Título:
Eye corners tracking for head movement estimation
Autor/es:
MARTÍNEZ, CÉSAR E.; GARCÍA CENA, C.E.; LARRAZABAL, A.J.
Lugar:
Budapest
Reunión:
Congreso; IEEE International Work Conference on Bioinspired Intelligence; 2019
Institución organizadora:
IEEE Hungary Section
Resumen:
Recently, video-oculographic gaze tracking has begunto be used in the diagnosis of a wide variety of neurologicaldiseases, such as Parkinson and Alzheimer. For this application,the so-called feature-based methods are used, more precisely, 2Dregression-based methods. They use geometrically derived eyefeatures from high-resolution eye images captured by zoominginto the user?s eyes. The main weakness of these methods is thatthe head of the user must remain motionless to avoid estimationerrors. In some patients, some involuntary movements cannot beavoided and it is necessary to measure them. In this paper, wetackle the measurement of head position as a way to improve thegaze tracking on these precision demanding medical applications.As a first stage, we propose to obtain the eye corners coordinatesas a reference point, since they are the most stable points infront of the eyeball and eyelids movements. The problem washandled as a regression problem using a coarse-to-fine cascadedconvolutional neural network in order to accurately regress thecoordinates of the eye corner. Particularly, with the aim ofachieving high precision we cascade two levels of convolutionalnetworks. Finally, we added temporal information to increaseaccuracy and decrease computation time. The accuracy of theestimation was calculated from the mean square error betweenthe predictions and the ground truth. Subjective performancewas also evaluated through video inspection. In both cases,satisfactory results were obtained.