Nanjing University of Posts and Telecommunicatins

Biyun Sheng

Jian Zhou

Fu Xiao

Hai Hu

Zhengxin Guo

Linqing Gui

Mingming Xu

Hui Cai

A static human pose estimation method based on channel state information (CSI) signal angle of arrival (AoA) estimation is disclosed. A signal angle of arrival in a sensing area is estimated from CSI and an image is constructed, and a human pose is estimated from the image using a teacher-student network. The method comprises the following specific steps: CSI signals in a sensing area are collected at different heights using a receive antenna column with a moving track. Two-dimensional AoA image features are constructed: CSI information is converted into one-dimensional AoA data by using a MUSIC algorithm, and the one-dimensional AoA data of different heights are combined into a two-dimensional AoA image, and an environmental denoise algorithm is designed. A teacher-student network model is constructed, and a student network based on the two-dimensional AoA image is supervised, wherein the student network model can independently estimate a human pose from CSI.

Static human pose estimation method based on CSI signal angle of arrival estimation

Growing community of inventors

Jiangsu, China

Biyun Sheng