Approaches for automatic low-dimensional human shape refinement with priors or generic cues using RGB-D data

作者：

Highlights：

• A novel and accurate method to refine low dimensional human shape using RGB-D data is proposed.

• Uses of multiple modalities do not carry any features from the shape provider.

• Combines low and high level observations jointly in multi-layer graph structure

• Extensive experiments showed that it outperforms compared suitable algorithms.

• Also, an existing method is extended by fusing more generic cues for this purpose.

摘要

•A novel and accurate method to refine low dimensional human shape using RGB-D data is proposed.•Uses of multiple modalities do not carry any features from the shape provider.•Combines low and high level observations jointly in multi-layer graph structure•Extensive experiments showed that it outperforms compared suitable algorithms.•Also, an existing method is extended by fusing more generic cues for this purpose.

论文关键词：Human shape refinement using RGB-D data,Multi-layer graph cut,Human body shape descriptor,Random decision forests,Refinement of low-dimensional representations,RGB-D

论文评审过程：Received 14 July 2014, Revised 10 May 2015, Accepted 11 May 2015, Available online 19 June 2015, Version of Record 27 June 2015.

论文官网地址：https://doi.org/10.1016/j.imavis.2015.05.001