Fast, Accurate Detection of 100,000 Object Classes on a Single Machine a prizewinning paper by Google Research scientists, describes a breakthrough in machine vision that can distinguish between a huge class of objects 20,000 times faster than before.
This so-called convolution operator is one of the key operations used in computer vision and, more broadly, all of signal processing. Unfortunately, it is computationally expensive and hence researchers use it sparingly or employ exotic SIMD hardware like GPUs and FPGAs to mitigate the computational cost. We turn things on their head by showing how one can use fast table lookup — a method called hashing — to trade time for space, replacing the computationally-expensive inner loop of the convolution operator — a sequence of multiplications and additions — required for performing millions of convolutions with a single table lookup.
We demonstrate the advantages of our approach by scaling object detection from the current state of the art involving several hundred or at most a few thousand of object categories to 100,000 categories requiring what would amount to more than a million convolutions. Moreover, our demonstration was carried out on a single commodity computer requiring only a few seconds for each image. The basic technology is used in several pieces of Google infrastructure and can be applied to problems outside of computer vision such as auditory signal processing.
Fast, Accurate Detection of 100,000 Object Classes on a Single Machine
(Image: Clutter, a Creative Commons Attribution Share-Alike (2.0) image from neofob's photostream)
Update: Whoops, David got there first! In a new paper published in the Proceedings of the Royal Society, researchers from UC Berkeley reveal that shoelace knots do not gradualy come loose, as was previously supposed — rather, they fail catastrophically and suddenly, thanks to strange and surprising stresses that they must endure.
Ad-blockers begat ad-blocker-blockers, which begat ad-blocker-blocker-blockers, with no end in sight.
Researchers from MIT, UC Berkeley, Lawrence Berkeley, and King Abdulaziz City for Science and Technology published a paper in Science describing a solar-powered device that uses a new type of metal organic framework (MOF) to extract up to three litres of water per day from even the most arid desert air.
Yeah, Bluetooth audio is pretty common these days, so why should you care about these earbuds? Look how happy that woman up above looks. She’s got FRESHeBUDS in. Boom. There’s your reason. She’s also at the beach and it appears to be a very nice day.But for the sake of promotion, wireless earbuds are fast becoming the […]
“Gets stuff done,” is a good way to be described by anybody. Especially by coworkers or bosses. Because whether you’re in finance or a children’s librarian, stuff needs to get done. But how do you make sure stuff gets done? You definitely can’t do all the stuff yourself, unless your company/organization/government office consists entirely of you. And […]
Even the most expensive pair of hi-fi headphones can’t match the feeling of bass rumbling through your body at a live show. That’s why music aficionados designed The Basslet, an accessory that reproduces that sensation from your wrist. Does it make your whole body shake with deep subs? Not really, because that would be terrifying, but […]