Towards High Performance Low Complexity Calibration in Appearance Based Gaze Estimation

• Published in: IEEE transactions on pattern analysis and machine intelligence Vol. 45; no. 1; pp. 1174 - 1188
• Main Authors: Chen, Zhaokang, Shi, Bertram E.
• Format: Journal Article
• Language: English
• Published: United States IEEE 01.01.2023; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Accuracy; Algorithms; Appearance-based gaze estimation; Bias; Calibration; Cameras; Color imagery; Complexity; Complexity theory; Datasets; Decomposition; deep neural networks; dilated convolutions; Estimation; eye tracking; Faces; Gaze tracking; Head movement; low complexity calibration; Magnetic heads; Model accuracy; NISLGaze dataset; Source code; State-of-the-art reviews; subject-dependent

Appearance-based gaze estimation from RGB images provides relatively unconstrained gaze tracking from commonly available hardware. The accuracy of subject-independent models is limited partly by small intra-subject and large inter-subject variations in appearance, and partly by a latent subject-depe...