Two mind-blowing TED talks share staggering research by computer vision researchers at the MIT Computer Science and Artificial Intelligence Laboratory: they have created algorithms capable of analyzing video to accurately reconstruct sound and other “invisible” information about objects. This allows them to hear unrecorded sounds and see invisible events, such as a person’s pulse and a baby’s gentle breathing.
The research is from the MIT Computer Science and Artificial Intelligence Laboratory. The technology allows objects in videos to be used as “microphones” by analyzing the tiny changes in the object’s position and color produced by sounds striking it.
In this first video, Michael Rubinstein discusses how sound can be extracted from objects depicted within silent videos, such as a bag of chips(!). In the second video, Abe Davis shares how information about the structure and properties of objects may also be deduced from video information – allowing computers to reconstruct video’d objects in virtual reality, so that they can be manipulated, understood and predicted.
The potential applications for surveillance and marketing are not insignificant. In We Know How You Feel, the New York Times’ Raffi Khatchadourian looks at how “Computers are learning to read emotion, and the business world can’t wait.” Will such technology make us more guarded and self-conscious, or slowly dissolve our attachments to privacy?