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)
Joe Karganis writes, "This is the 'Co-Assignment Galaxy' created by David McClure. It maps the top 160K titles in the new Open Syllabus 2.0 dataset, based on the frequency with which those texts are assigned (reflected in the size of the dot) and assigned together (reflected in the location and clustering of the dots). It's […]
If you only look at porn with your browser in incognito mode, your browser will not record your porn-viewing history; but the porn sites themselves overwhelmingly embed tracking scripts from Google and Facebook in every page: 93% of 22,484 porn sites analyzed in a New Media & Society paper had some kind of third-party tracker, […]
Ed Hawkins, a climate scientist, created Show Your Stripes as a way to easily visualize the past century's climate change: give it a location and it will render a series of stripes representing a century's worth of average annual temperatures (above: global average temperature); as Kottke notes: "The warming patterns for particular regions are not […]
Vape technology has been around long enough that vapers are starting to get picky about their gear. Luckily, so are we. From disposable models to cutting-edge touchscreen atomizers, there’s a vaporizer in this roundup to suit every taste. Hera 2 – World’s Most Advanced Dual-Use Vaporizer Choose between dry herb or oil extraction modes – […]
With enough practice and commitment, anyone can be a visual artist. But without the right instruction, that time spent honing your skills could seem like an eternity. If you really want to see where your talent can take you, you need sound fundamentals – and no matter what discipline or genre you lean toward, the […]
Theoretically, there’s never been an easier time for marketers. The ubiquity of social media means a good word – or a good brand – can spread like wildfire with very little effort. But as limitless as the internet is, there’s a lot of competition and noise to contend with. And the vast graveyard of failed […]