{"id":1357,"date":"2021-10-13T07:50:14","date_gmt":"2021-10-13T07:50:14","guid":{"rendered":"https:\/\/sail.usc.edu:\/~ccmi\/?p=1357"},"modified":"2021-10-16T08:26:48","modified_gmt":"2021-10-16T08:26:48","slug":"active-speaker-detection","status":"publish","type":"post","link":"https:\/\/sail.usc.edu:\/ccmi\/active-speaker-detection\/","title":{"rendered":"Active speaker detection"},"content":{"rendered":"\n
\n

Project description<\/h2>\n\n\n\n

“When<\/em>” and “where<\/em>” are the fundamental pillars of Computational media intelligence, for developing a holistic understanding of a scene which translates to locate the action of interest in the space and time. In this project, we develop a system to automatically detect the active speakers in space and time, specifically for media content.<\/p>\n<\/div><\/div>\n\n\n\n

\n

Methodology<\/h2>\n<\/div><\/div>\n\n\n\n
\n

This work is inspired by our preliminary work [1], where we presented a cross-modal system for the task of voice activity detection by observing just the visual frames. We performed a thorough analysis and showed that the learned embeddings can locate the human bodies and faces. The cross-modal architecture is shown below: <\/p>\n<\/div><\/div>\n\n\n\n

\"\"<\/figure>\n\n\n\n
\n

Demonstration of the system output.<\/h2>\n<\/div><\/div>\n\n\n\n
\n
\n
\n