This paper is a summary of the 2007 CLEAR Evaluation on the Classification of Events, Activities, and Relationships which took place in early 2007 and culminated with a two-day workshop held in May 2007. CLEAR is an international effort to evaluate systems for the perception of people, their activities, and interactions. In its second year, CLEAR has developed a following from the computer vision and speech communities, spawning a more multimodal perspective of research evaluation. This paper describes the evaluation tasks, including metrics and databases used, and discusses the results achieved. The CLEAR 2007 tasks comprise person, face, and vehicle tracking, head pose estimation, as well as acoustic scene analysis. These include subtasks performed in the visual, acoustic and audio-visual domains for meeting room and surveillance data.
Conference Dates: May 8-9, 2007
Conference Location: Baltimore, MD
Conference Title: Multimodal Technologies for Perception of Humans: Joint Proceedings of the CLEAR 2007 and RT 2007 Evaluation Workshop
Pub Type: Conferences
computer vision, human language technology, evaluation infrastructure