Google have previously taken out a patent that would see the Mountain View-based company capable of identifying faces in YouTube videos, but now they've gone a step further by securing a patent that would detect objects in a video.
The objects can be anything from a pole, to a sidewalk, or a car, or even a building. Google, instead of asking the creator to label objects each time, proposes using a database of "feature vectors" that would include color, movement, shape and texture, which would automatically identify subjects in the frame using these trait.
The tagging of items in videos would mostly be completed by those uploading videos to YouTube, so Google wouldn't even have to do that much work. After a while, there should be enough data in their database to keep things going very well.
Further Reading: Read and find more Business, Financial & Legal news at our Business, Financial & Legal news index page.