Google’s ‘Deep Stereo’ could transform Street View into an immersive digital world

Google researchers have designed an algorithm that can transform the still images on Street View into fluid video – a step that could one day make the mapping service a far more realistic experience.

The software, which allows a close-up view of many city streets, tourist attractions and residential suburbs in the developed world, currently features individual panoramic images, allowing the user take a close look at a location but not move seamlessly around it.

Now Google researchers have developed what they call the Deep Stereo code – which automatically generates a third image to fill in the “missing frames”, according to the MIT Technology Review.

The service guesses what the frame would “likely” show by studying the two existing images from any scene.

The code works best in plain surroundings like an unchanging highway, however it becomes trickier in art galleries and shopping malls where the details are finer.

Google’s John Flynn told MIT that an “exact solution” was not possible.

“An exact solution would require full 3-D knowledge of all visible geometry in the unseen view,” he said.

However, the results are still impressive.