A Faster R-CNN and recurrent neural network based approach of gait recognition with and without carried objects

作者：

Highlights：

• Proposing a novel method of gait recognition both with and without carried objects.

• The faster region convolutional neural network detects the person in each frame.

• Feature vector of the walking pattern is generated using convolutional operations.

• Feature vectors are studied using two different models of recurrent neural network.

• The best recognition accuracy obtained is 98.54% among four datasets.

摘要

•Proposing a novel method of gait recognition both with and without carried objects.•The faster region convolutional neural network detects the person in each frame.•Feature vector of the walking pattern is generated using convolutional operations.•Feature vectors are studied using two different models of recurrent neural network.•The best recognition accuracy obtained is 98.54% among four datasets.

论文关键词：Gait recognition,With and without CO,Faster R-CNN,RNN,LSTM,BLSTM

论文评审过程：Received 29 July 2021, Revised 10 May 2022, Accepted 31 May 2022, Available online 7 June 2022, Version of Record 12 June 2022.

论文官网地址：https://doi.org/10.1016/j.eswa.2022.117730