ACN: Occlusion-tolerant face alignment by attentional combination of heterogeneous regression networks
作者：
Highlights：
• We propose an occlusion-tolerant highly accurate face alignment method.
• It combines a coordinate and a heatmap regression network with a spatial attention.
• It compensates complementarily overall fitting tendency and detailed localization.
• It uses coordinate-to-heatmap and heatmap-to-coordinate conversion networks.
• It achieves state-of-the-art accuracy In experiments on several benchmarks.
摘要
•We propose an occlusion-tolerant highly accurate face alignment method.•It combines a coordinate and a heatmap regression network with a spatial attention.•It compensates complementarily overall fitting tendency and detailed localization.•It uses coordinate-to-heatmap and heatmap-to-coordinate conversion networks.•It achieves state-of-the-art accuracy In experiments on several benchmarks.
论文关键词：Face alignment,Facial landmark localization,Attentional combination network,Converting networks
论文评审过程：Received 17 February 2020, Revised 26 October 2020, Accepted 25 November 2020, Available online 2 December 2020, Version of Record 2 March 2021.
论文官网地址：https://doi.org/10.1016/j.patcog.2020.107761
