Diagnosing deep learning models for high accuracy age estimation from a single image

作者：

Highlights：

• We comprehensively diagnose deep learning models for the age estimation problem.

• We investigate three problem formulations with five loss functions.

• We propose a multi-task architecture for age, gender and race classification.

• We obtain state-of-the-arts age estimation results on two large benchmark datasets.

摘要

Highlights•We comprehensively diagnose deep learning models for the age estimation problem.•We investigate three problem formulations with five loss functions.•We propose a multi-task architecture for age, gender and race classification.•We obtain state-of-the-arts age estimation results on two large benchmark datasets.

论文关键词：Age estimation,Deep learning,Multi-task learning

论文评审过程：Received 14 July 2016, Revised 23 November 2016, Accepted 4 January 2017, Available online 5 January 2017, Version of Record 12 March 2017.

论文官网地址：https://doi.org/10.1016/j.patcog.2017.01.005