Important object detection from TV programs based on production technique estimation
Tomohiko Takahashi, Masaru Sugano, Shigeyuki Sakazawa · 2010
This paper proposes a new method to automatically detect important objects from video content. Our approach extracts video frames that show important objects, by recognition of typical TV production techniques. We also present a prototype application for DVRs, which utilizes the proposed method to enable easy access to important scenes of TV programs.