Exploiting scene maps and spatial relationships in quasi-static scenes for video face clustering
作者：
Highlights：
• We propose a video face clustering algorithm for quasi-static scene (QSS) videos.
• QSS videos include e.g., talk shows, TV debates, TV games and symphonic concerts.
• A map of the scene and the spatial relationships between people are inferred.
• By using them, we match faces in crowded shots with lack of visual detail.
• We show that spatial information is the missing piece for effective face clustering.
摘要
•We propose a video face clustering algorithm for quasi-static scene (QSS) videos.•QSS videos include e.g., talk shows, TV debates, TV games and symphonic concerts.•A map of the scene and the spatial relationships between people are inferred.•By using them, we match faces in crowded shots with lack of visual detail.•We show that spatial information is the missing piece for effective face clustering.
论文关键词：Video face annotation,Face clustering,Re-identification
论文评审过程：Received 24 August 2015, Revised 14 September 2016, Accepted 8 November 2016, Available online 23 November 2016, Version of Record 4 December 2016.
论文官网地址：https://doi.org/10.1016/j.imavis.2016.11.005
