An efficient multitask neural network for face alignment, head pose estimation and face tracking

作者：

Highlights：

• Employ the structural information at shallow layer for efficient face alignment.

• Generate a heatmap to provide the geometric and attention information for network.

• Propose an efficient multitask framework for video-based processing.

• Conduct comparison experiments and ablation studies on various datasets.

摘要

•Employ the structural information at shallow layer for efficient face alignment.•Generate a heatmap to provide the geometric and attention information for network.•Propose an efficient multitask framework for video-based processing.•Conduct comparison experiments and ablation studies on various datasets.

论文关键词：Face alignment,Head pose estimation,Face tracking,Multitask neural network

论文评审过程：Received 19 August 2021, Revised 25 January 2022, Accepted 25 April 2022, Available online 19 May 2022, Version of Record 3 June 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.117368