This AI tool can convert a blurry photo into high-quality portrait shot

Researchers at the US-based Duke University have come up with a new Artificial Intelligence (AI) tool that can process a blurry low-resolution image into a high-quality portrait with richer details. Researchers have named this AI tool as PULSE, which refers to “Self-Supervised Photo Upsampling via Latent Space Exploration of Generative Models.”

It is worth noting that the AI-based image processing tools have been around for quite some time. What sets the Pulse apart from the rest is its ability to enhance a handful of pixels into a realistic portrait, bumping up the resolution by up to 64 times.

How it works

The Duke researchers adopted a different method for image enhancement. Instead of letting the AI guess what extra pixel is necessary and combining them with existing images, Duke researchers’ tool looks for AI-generated high-quality portrait shots and delivers an image that looks closest to the input image.

Researchers used a Machine Learning tool known as “Generating Adversarial network” (GAN). The tool is essentially a combination of two neural networks fed with the same set of data sets (photos in this case). The first network is responsible for generating AI-created portraits. The second network handles the output.

ALSO READ: Facebook can now convert your old 2D photos into 3D

"The system can convert a 16x16-pixel image of a face to 1024 x 1024 pixels in a few seconds, adding more than a million pixels, akin to HD resolution. Details such as pores, wrinkles, and wisps of hair that are imperceptible in the low-res photos become crisp and clear in the computer-generated versions," claims the report.

Use cases and possible misuse

According to researchers, the Pulse AI can be used for converting any low-resolution shots into high-quality, sharp and more realistic images. The tool can be used for microscopy and even in fields of satellite imagery and astronomy.

On possible misuse of the AI, researchers clarified that the tool isn’t meant to identify real people. For instance, it won’t convert blurry photos of people taken from a security camera. Instead, the AI delivers the faces that look real but are AI-generated.

You can learn more about new research here.