Body-movement-based human identification using convolutional neural network

作者：

Highlights：

• Our method is body movement-based human identification using front and back view.

• Our identification method is robust to the cases of people carrying items or with various poses.

• Deep learning method using thermal difference image-based three-channel inputs is used.

• Our collected database and trained CNN are public to other researchers.

摘要

•Our method is body movement-based human identification using front and back view.•Our identification method is robust to the cases of people carrying items or with various poses.•Deep learning method using thermal difference image-based three-channel inputs is used.•Our collected database and trained CNN are public to other researchers.

论文关键词：Low-illumination corridor,The front and back view of body,Human identification,Convolutional neural network

论文评审过程：Received 12 October 2017, Revised 10 January 2018, Accepted 8 February 2018, Available online 9 February 2018, Version of Record 15 February 2018.

论文官网地址：https://doi.org/10.1016/j.eswa.2018.02.016