The days of trying to remember when and where you took a picture could be over thanks to a machine learning technology being developed by Qualcomm.

The tech, currently referred to as Zeroth, has the ability to scan images you've taken and match points of reference on them, whether they are of landscapes, people or even horses.

The company hopes that software developers will take the technology and build it into their photo apps.

While searching and finding images on your phone or computer is one thing, the company has already taken the technology and applied it to live capture as well.

Similar to intelligent auto modes on compact cameras, but "more advanced" as one Qualcomm spokesperson told us, the technology is capable of monitoring what is in the picture and automatically changing the settings accordingly.


However, unlike current auto modes that mostly rely on light and dark histogram data to evaluate a scene, Qualcomm's system detects the objects in sight based on what it has been taught. And the more pictures you take the more it learns.

Qualcomm told Pocket-lint that this data can then be applied to people, in particular people that it knows. In a demo of the new tech at the company's stand at Mobile World Congress, people already programmed into the system had markers telling us who they were.


But where Qualcomm says the technology has real potential is in capturing handwriting and turning it into text. Because the system is able to learn, Qualcomm has already started to program it with multiple styles of handwriting to help it get better at learning people's scrawl.

The new technology is expected to slowly start appearing in our phones and tablets over the next couple of years.