Face analysis through semantic face segmentation

作者：

Highlights：

• A multi-feature framework performing face analysis, including facial part segmentation, head pose estimation, gender recognition, and expression classification.

• Performance of the proposed face analysis framework is evaluated on four standard face databases, namely Pointing’04, FEI, FERET, and MPI, with results which outperform the current state-of-the-art.

• A public face data repository, namely FASSEG, containing more than 270 images taken from the MIT-CBCL, Pointing’04, and the FEI databases annotated pixel-wise on six semantic classes.

摘要

•A multi-feature framework performing face analysis, including facial part segmentation, head pose estimation, gender recognition, and expression classification.•Performance of the proposed face analysis framework is evaluated on four standard face databases, namely Pointing’04, FEI, FERET, and MPI, with results which outperform the current state-of-the-art.•A public face data repository, namely FASSEG, containing more than 270 images taken from the MIT-CBCL, Pointing’04, and the FEI databases annotated pixel-wise on six semantic classes.

论文关键词：Face segmentation,Head pose estimation,Gender recognition,Face expression classification

论文评审过程：Received 12 April 2018, Revised 16 November 2018, Accepted 12 January 2019, Available online 18 January 2019, Version of Record 31 January 2019.

论文官网地址：https://doi.org/10.1016/j.image.2019.01.005