This tool uses neural networks to improve images with poor quality

Neural networks are one of the main pillars of artificial intelligence. They are based on applying 'human' attitudes and responses to programming, getting certain situations resolved in a way similar to how a person's brain would ... or at least approaching it. It has more and more applications, and one of them is the improvement of images that do not look as they should.

Artificial Intelligence has more and more expressions, and one of them is the improvement of the quality of images. This is the case of the Let's Enhance project, carried out by a group of Ukrainian developers, using neural networks to improve images that are blurred or have low resolution.

Your system has memorized the habitual patterns in certain types of images: faces, portraits, group photos, landscapes, urban environments, vehicles ... Based on them, it detects what exactly has the image that we ask you to improve, and sets to to work.

A Let's Enhance we can not ask for superhuman wonders as if we were in front of the CSI computer, able to obtain clear portraits of the perpetrator of the crime reflected in a screw of a license plate of a car at the bottom of the scene. What we can expect from this tool is that it solves the resolution problems of photographs that we have stored for the good memory that we are supposed to have, but not for its quality. Or those, for example, in which we did not take all the necessary care at the time of shooting.

It also comes in handy for a particular use: getting photographs from small fragments of other photographs, that is, making cuts and getting virtually no loss of quality.

The process is simple: 'drag and drop' and wait. Above all, wait: the program can take two minutes to hours. Everything will depend on the load volume of the platform and the complexity of the image in question.

When you have analyzed and 'repaired' the image, it returns it in three different modes:

- Anti-JPEG: it is limited to eliminate the usual artifacts that appear when an image is compressed with this format.

- Boring: enlarges the image and details it, but in a moderate way so that it can not be evident that there has been an edition.

- Magic: the most aggressive editing mode, for when we can afford much more impressive results.

Let's Enhance is in its first stages of publication, so its creators have already warned that many improvements will be coming to the tool during the coming months. At the moment it is a good option to have stored in our favorites for when we need your services ... but not to solve crimes as CSI.