Apple has acquired Spektral, a Danish start-up specializing in computer vision, for augmented reality technology – TechCrunch



[ad_1]

In the wake of Apple's signing this morning of a $ 600 million contract for the acquisition of intellectual property, talent and licenses for Dialog Semiconductor in Europe, the group also confirmed the new acquisition of a smaller start-up in the region.

Apple has purchased Spektral, a Denmark-based computer vision company, which uses segmentation technology, a more efficient way of "slicing" the numbers from their antecedents into digital images, for an amount of about 30 million. dollars.

This type of technology can be used, for example, in augmented reality, but also for more standard applications such as school photo (which was actually the first market targeted by the startup in 2015).

Rumors about the deal began to surface yesterday, first in the Danish financial newspaper Børsen. We contacted Apple and the company has now confirmed the contract with its standard statement: "Apple occasionally buys smaller technology companies, and we do not usually discuss our goals or our projects."

From what we understand, the acquisition took place some time ago – which corresponds to a LinkedIn profile for Toke Jansen, co-founder of Spektral, but he now notes that in December In 2017, he was responsible for computer imaging at Apple. .

Other people associated with the company, including Henrik Paltoft, his co-founder, have not updated their profile. It is therefore difficult to know how many other people have joined the group. Børsen reports that the agreement includes the company's engineers and amounted to around DKK 200 million, which equates to around USD 31 million.

Spektral started life as CloudCutout, based on algorithms from Jansen's doctorate. The start-up initially introduced its product as a cheaper and more efficient "green screen" technology, allowing primary images to be removed from their single (usually green) backgrounds or standard designs, the first version of the product being constructed by forming the system to more than 100,000 professional cuts.

Spectral's first app may have been the rather retro world of school images, but what's most noticeable here is what Spektral could contribute to Apple's imaging business. This is not just about apps that Apple has not yet launched, but also about improving the quality of those already on the market, from legacy products such as PhotoBooth to ARKit, the mobile development platform of the society.

Segmentation can help to add live filters to human figures in a photo, but can also be effective in obstructing RA environments behind the figures so that the digital content of the RAs appears interacting with the position of humans.

Spektral's segmentation technology is also capable of running on mobile phones, potentially making it a faster and more efficient way to process RA images directly on devices.

"To provide high quality cutouts, the heart of our engine exploits recent advances in spectral graph theory and neural networks. The calculation of pixel transparencies (alpha channel) for a single image involves the resolution of several large-scale equation systems, as well as the realization of outside multiple anticipation passes in our neural networks, "said the founders at the launch of the startup.

"We pose the problem of determining an alpha channel of an image as an automatic learning task. Compared to the usual chroma key, this allows us to consider a much wider range of backgrounds since the model will learn, that is, texture representations from existing training data. "

Computer vision is a key goal (sorry!) For Apple for quite some time now. The iPhone and Mac giants have made more than 40 acquisitions in Europe over the past 10 years – I guess we may still have some way to go – and a number of its acquisitions, both in Europe and elsewhere , were performed in the field of computer vision. Akonia Holographics, InVisage Technologies, Regaind, Vrvana, SensoMotoric Instruments, Indoor.io, Flyby Media, Emotient, Faceshift, Metaio, Polar Rose and much more.

Additional report Natasha Lomas.

[ad_2]
Source link