COVID-19 Cases

Blur faces are to be enhanced in super resolution images by AI


Researchers found a way to use artificial intelligence to transform a few dozen pixels into a high resolution face image. A Duke University team in the US created an algorithm that can 'imagine' realistic faces of people with eight times more efficiency than earlier methods. "There was never a super resolution

With help of AI we can easily convert low resolution photos to high resolution

The AI images do not look like real people, instead they are plausibly real faces. This means that persons from low-resolution images captured by security cameras can not be identified.

The PULSE system of Dr Rudin and its team creates pictures with 64 times the resolution of the original fluctuating image (Photo Upsample via Latent Space Exploration).

By reverse engineering the image of high-resolution images that look like a low-resolution image, the PULSE algorithm is capable of achieves such a high resolution.

Through the process, the facial characteristics such as eyelashes, teeth and wrinkles are recognizable and detailed in the low-resolution image.

"PULSE is looking for images that are down to the original low resolution image rather than starting with the low resolution, and slowly adding detail," says a paper that details the research. In theory, the system can be used for almost anything from medication to microscopy, astronomy and satellite image imagery in low resolution pictures.

This means that high pitched, poor quality images can be imagined in high resolution from far away planets and solar systems. The research will be presented at the CVPR this week at the 2020 CVPR Conference.

Article Edited by | Jhon H |

we hope you're doing well if you see any inappropriate phrases please let us know on our contact page at the bottom. thank you! .