Google patented automatic labeling of objects in a video
Google patented a system for object Recognition that would label all the elements of a video, either inert or living things. The idea is that instead of asking the creator of a video that labels the objects shown, Google intends to use a database ‘feature vectors’ such as color, motion, shape and texture.
The system would predict which visual objects corresponding to these vectors in order to automatically assign a name, and have a sufficient level of detail to distinguish moving objects through features in common with other known elements.
Now we just have to know how Google will use this new system, but it would make sense to give YouTube’s property automatically tag all those videos of pets that abound in your website.Google, Image Recognition, patent, Patents, Recognition, recognition software