Anatomical-plane-based representation for human–human interactions analysis
作者：
Highlights：
• We propose a view-invariant motion-pose geometric descriptor (MPGD) as a representation of human–human interaction (HHI).
• The MPGD utilizes the concept of anatomical planes to model the interacting humans.
• The MPGD can capture the semantic meaning of body-parts for two interacting humans.
• Using the MPGD, we developed frameworks to analyze HHI from two perspectives: classification and prediction from an input video.
摘要
Highlights•We propose a view-invariant motion-pose geometric descriptor (MPGD) as a representation of human–human interaction (HHI).•The MPGD utilizes the concept of anatomical planes to model the interacting humans.•The MPGD can capture the semantic meaning of body-parts for two interacting humans.•Using the MPGD, we developed frameworks to analyze HHI from two perspectives: classification and prediction from an input video.
论文关键词：Human–human interaction classification,Human–human interaction prediction,Bag of words,Motion-pose geometric descriptor
论文评审过程：Received 2 November 2013, Revised 19 February 2015, Accepted 3 March 2015, Available online 11 March 2015.
论文官网地址：https://doi.org/10.1016/j.patcog.2015.03.002
