TAA-GCN: A temporally aware Adaptive Graph Convolutional Network for age estimation

作者：

Highlights：

• Proposed a new age estimation algorithm using Facial and Skeletal-Cosmetic graphs.

• The proposed TAA-GCN includes two new modules AGCL and TMM.

• The TMM capture non-ordinal temporal dependencies and the AGCL refines graphs.

• Proposed a new graph representation based on face, skeleton, posture, and clothing.

• The proposed algorithm outperformed SOTA methods on four public datasets.

摘要

•Proposed a new age estimation algorithm using Facial and Skeletal-Cosmetic graphs.•The proposed TAA-GCN includes two new modules AGCL and TMM.•The TMM capture non-ordinal temporal dependencies and the AGCL refines graphs.•Proposed a new graph representation based on face, skeleton, posture, and clothing.•The proposed algorithm outperformed SOTA methods on four public datasets.

论文关键词：Age estimation,Graph convolutional network,Facial graphs,Skeletal graphs

论文评审过程：Received 12 April 2022, Revised 22 August 2022, Accepted 20 September 2022, Available online 23 September 2022, Version of Record 29 September 2022.

论文官网地址：https://doi.org/10.1016/j.patcog.2022.109066