This sweater is like the invisibility cloak from Harry Potter but with a twist | The Financial Express

This sweater is like the invisibility cloak from Harry Potter but with a twist

The University of Maryland has several achievements worth mentioning under its name.

This sweater is like the invisibility cloak from Harry Potter but with a twist
This sweater is like the invisibility cloak from Harry Potter but with a twist. (Photo Credits: Computer Science University Maryland website)

What if we told you that Harry Potter’s one of the fabled deathly hallows and desired by many – the “cloak of invisibility” is real and it can help you trick modern face detection mechanisms? This isn’t some prank or joke. Researchers at the University of Maryland along with Facebook Artificial Intelligence have actually developed an invisibility cloak which is a pullover and it will hide your face from modern AI cameras.

While the ultimate goal of the study was to find vulnerabilities in the machine learning systems, it resulted in a unique print on clothes that AI cameras can’t see.  

The sweater features a stay-dry microfleece lining and uses adversarial patterns to trick AI cameras into detecting the wearer.  

A Gagagadget.com report tries to explain the process stating that the experts used the SOCO dataset on which the computer vision algorithm YOLOv2 is trained, identified a pattern that helps to recognize a person, created an opposite pattern and transformed it into an image – a print on a sweater. As a result, the owner of such a sweater can hide from detection systems.

“This stylish pullover is a great way to stay warm this winter, whether in the office or on the go. It features a stay-dry microfleece lining, a modern fit, and adversarial patterns that evade the most common object detectors. In this demonstration, the YOLOv2 detector is evaded using a pattern trained on the COCO dataset with a carefully constructed objective,” wrote one of the researchers on the official page of Computer Science, University of Maryland.  

Describing the approach used for creating the sweater, a researcher writes that the team loads images from the COCO detection dataset, and passes them through a detector. When a person is detected, a pattern is rendered over that person with random perspective, brightness, and contrast deformations. A gradient descent algorithm is then used to find the pattern that minimizes the “objectness scores” (confidence in the presence of an object) for every object prior.

The University of Maryland has several achievements worth mentioning under its name. Its researchers recently developed a sensor made from a futuristic quantum material that can detect food freshness to help grocery stores and production facilities slash waste.

ALSO READ | Intel just made a real-time deepfake detector, claims it has whopping 96% accuracy

Get live Share Market updates and latest India News and business news on Financial Express. Download Financial Express App for latest business news.

First published on: 21-11-2022 at 14:14 IST