Parsing human image by fusing semantic and spatial features: A deep learning approach

作者：

Highlights：

• Human image parsing based on the context semantic and spatial detail feature fusion.

• Context and spatial pyramid unit encode semantic and spatial content accordingly.

• Dual pyramid unit conduces to parsing performance from human image to body label.

• The proposed approach strikes a balance between the speed and parsing performance.

摘要

•Human image parsing based on the context semantic and spatial detail feature fusion.•Context and spatial pyramid unit encode semantic and spatial content accordingly.•Dual pyramid unit conduces to parsing performance from human image to body label.•The proposed approach strikes a balance between the speed and parsing performance.

论文关键词：Human image parsing,Features fusion,Convolutional neural network,Multi-Level map

论文评审过程：Received 1 March 2020, Revised 13 May 2020, Accepted 18 May 2020, Available online 28 May 2020, Version of Record 28 May 2020.

论文官网地址：https://doi.org/10.1016/j.ipm.2020.102306