FEANet: Foreground-edge-aware network with DenseASPOC for human parsing

作者：

Highlights：

• Use foreground guidance to globally reducing the effect of pixels occupied by non-human object parts.

• Use edge guidance to locally preserve the boundaries of human parts.

• Embed DenseASPOC module to extracte semantic features of multi-scale objects.

• Combine geometric context with non-locally object-based context knowledge.

摘要

•Use foreground guidance to globally reducing the effect of pixels occupied by non-human object parts.•Use edge guidance to locally preserve the boundaries of human parts.•Embed DenseASPOC module to extracte semantic features of multi-scale objects.•Combine geometric context with non-locally object-based context knowledge.

论文关键词：Human parsing,Semantic segmentation,Foreground-edge awareness,Non-local operation

论文评审过程：Received 22 January 2021, Revised 23 February 2021, Accepted 24 February 2021, Available online 1 March 2021, Version of Record 6 March 2021.

论文官网地址：https://doi.org/10.1016/j.imavis.2021.104145