Hybrid eye center localization using cascaded regression and hand-crafted model fitting

作者：

Highlights：

• Eye centers are detected using a cascade of regression trees with HoG features.

• Robust circle fitting to iris contour is used to refine the eye center locations.

• Hand-crafted detector is used to train the regressor without manual annotations.

• State-of-the-art eye center localization results are achieved.

摘要

•Eye centers are detected using a cascade of regression trees with HoG features.•Robust circle fitting to iris contour is used to refine the eye center locations.•Hand-crafted detector is used to train the regressor without manual annotations.•State-of-the-art eye center localization results are achieved.

论文关键词：Eye center detection,Eye tracking,Facial feature alignment,Cascaded regression

论文评审过程：Received 23 June 2017, Revised 8 January 2018, Accepted 10 January 2018, Available online 31 January 2018, Version of Record 8 February 2018.

论文官网地址：https://doi.org/10.1016/j.imavis.2018.01.003