Sachin Farfade and Mohammad Saberian at Yahoo Labs in California and Li-Jia Li at Stanford University, revealed an algorithm that can spot faces at an angle, even when partially occluded. The ability to spot faces from any angle, and even when partially occluded, has always been a uniquely human capability. But now the researchers say that their work will revolutionize the image search engines.

As MIT's Technology Review reports, by taking advantage of the advances made in recent years on a type of machine learning known as a deep convolutional neural network, the researchers have managed to train a many-layered neural network using a vast database of annotated examples. These examples were pictures of faces from many angles and orientations, and also millions of images without faces. They then trained their neural net in batches of images over thousand of iterations.

The result is a single algorithm that can spot faces from a wide range of angles, even when partially occluded. And it can spot many faces in the same image with remarkable accuracy.

What’s more, their algorithm is significantly better at spotting faces when upside down, something other approaches haven’t perfected.

The great promise of this kind of algorithm is in image search. It is a step toward achiving the detection of images taken of specific people. And of course, the technology could be applied to digitized images including vast stores of video and CCTV footage.