Google will recognize the objects in the video YouTube
Recently we reported about experiments conducted by Google in its laboratories and were focused on the construction of artificial neural networks for automated processing of digital content imitating as closely as possible, the processes performed by the human brain . The platform set up by the technicians of the Mountain View giant has been able to recognize pictures of cats present in the video posted on YouTube without the neural network had previously been educated on the identity of felines.
Along the same wavelength there is the recording of a patent that was just granted to Google by the competent office in the United States. With “ Automatic large scale video object recognition ”, the engineers of the company founded by duo Larry Page Sergey Brin-have described a new algorithm that will automatically recognize objects (and possibly people, add) present in the video , such as those make up the immense archive of YouTube. Employing the system detailed in this document stamped by the U.S. Patent Office, Google can examine the contents of each image that makes each video, go in search of particular objects and draw up a sort of catalog of such elements . Suppose you have uploaded to YouTube a video that shows, along with some friends at the Spanish Steps in Rome.
Google, using its patented technology, analyze the entire contents of the movie clip and some will match label (“tag”) to the video: surely specify, at least, “ Rome ”,” Spanish Steps“and” Spanish Steps ” the church overlooking the famous staircase. Grande weight assume however the way in which Google decides to implement the new algorithm: the technicians of the company will have to move with leaden feet to avoid triggering the ire of safeguarding privacy.