Facebook is creation large strides in using a synthetic comprehension systems for picture recognition, though it’s also operative on bringing this record to video. As a association demoed during a F8 developer discussion in San Francisco today, it has a group operative on automatically tagging people in videos.
As Facebook’s executive of appurtenance learning Joaquin Quiñonero Candela pronounced during his keynote today, a thought here is to concede we to hunt for people in any of a videos they have common with you. Say we are on a live video feed with a crony and another crony walks into a video and has a brief review with you. Typically, that would be a really fleeting knowledge since it would be tough to find this impulse again.
Soon, Facebook might be indexing this impulse automatically for we and will let we find it again simply by acid for your friend’s name. You could afterwards burst right into a video during accurately a impulse your crony walks in.
Facebook is also operative on automatically captioning videos, and it is fathomable that it could use a same image-recognition techniques for video that it uses for detecting objects in still images.