American sign language recognition and training method with recurrent neural network

作者：

Highlights：

• An American Sign Language recognition model was developed using Leap Motion.

• LSTM-RNN with kNN method was proposed for recognition 26 alphabets.

• 3D motion of hand gesture and relevant 30 features were extracted.

• 26 alphabets with recognition rate of 99.44% accuracy was obtained.

摘要

•An American Sign Language recognition model was developed using Leap Motion.•LSTM-RNN with kNN method was proposed for recognition 26 alphabets.•3D motion of hand gesture and relevant 30 features were extracted.•26 alphabets with recognition rate of 99.44% accuracy was obtained.

论文关键词：American sign language,Leap motion controller,Learning application,Sign recognition system

论文评审过程：Received 22 October 2019, Revised 3 August 2020, Accepted 27 November 2020, Available online 3 December 2020, Version of Record 8 December 2020.

论文官网地址：https://doi.org/10.1016/j.eswa.2020.114403