There is no end in how many photos we produce. In fact, all the projections show that we have only just started and that the 880 billion photos we will take in 2014 are nothing compare to what is to come. But who organises all these images? Enter Euvision Technologies , from the University of Amsterdam, who promises to let you continue shooting carelessly while it takes care of automatically classifying your precious moments. Or any moment. In fact, it can classify any image you submit into whatever category you choose as long as you teach it . As a proof of concept, the company released a simple version of their technology as a phone app under the name Impala. While useful, it is a far cry from their full potential. We caught up with Harro Stokman, Founder and Ceo of Euvision to learn more (after the short intro):
Euvision Technologies: Our Story from Euvision Technologies on Vimeo.
Tell us more about Impala. What does it do and why did you create it ?
Impala looks at your pictures and determines what’s visible in it. More precisely: it calculates a series of probability scores per picture. So “the probability that there is a cat on the picture is 99% and the probability that it is outdoor is 1%”. The core technology originates from academic research from the University of Amsterdam.
Given the tsunami of pictures on social media, I thought it was a waste to let this kind of technology collect dust on academic shelves, so I took it off the academic shelves and brought it to the market.
What makes your product unique ?
Well, there are a few ways to determine what’s on a picture. First way is of course ask a human being to describe it, which is not very scalable, or fun to do. Secondly, you can try to analyze the text around pictures, which works for images on webpages on the Internet, but not for images on my phone. Thirdly you can match against databases of pictures and see if the picture you are analyzing is already in your database, but this only works for known pictures (ie pictures of wine bottles in your database). Finally you can try to make a model of what you’d like to find in your picture (like biometrical models for faces, or models of characters on a license plate). We don’t do any of this. Instead we follow a completely different approach. Our technology is trainable to detect anything. So if we’d like to detect sunsets, we feed the system with pictures of sunsets and our software Impala teaches itself to recognize sunsets. Or Nike sneakers, or sexual child abuse, or bombvests or whatever else you’d like to find. The big advantage of this is that we can classify all pictures for any type of search.
How big is the team today ?
We have a highly focussed team of 8 people. All PhD’s, except for the sales guy.
Have you raised any funds and if so, how much ?
No, we are proud that we are bootstrapped, and profitable. We only had seed capital, which was provided by the management team.
What is the long-term vision of your company, release more powerful apps or be a solution provider?
We believe that reaping the hidden value in pictures has only just begun and we feel we are well positioned to facilitate this trend. We simply aim to have our technology process as many videos and images as possible. Therefore we team up with customers and software vendors who work with vast amounts of pictures. The Impala app was just a technology showcase for us to attract attention and that worked well.
Would you ever see Euvison Technologies creating a Flickr competing service based on your auto sorting technology ?
We’d rather help Flickr further improve their services than compete with them.
Who is your typical client today and what do they do with your solution ?
We have a variety of clients, ranging from dating sites, social media platform owners to law enforcement. They use Impala to perform automatic classification of pictures (as opposed to doing that manually). In social media, profile pictures are moderated using Impala which saves hundreds of workers in low wage countries). In Law Enforcement we support the fight against Child Abuse Materials (an area where we’d love to support Twitter also, they do have an issue in this field it seems). And a new application that one of our clients developing now is auto filling photobooks. So Impala classifies pictures and this classification is used to suggest the best set of images for your photobook.
Your auto sorting works with object recognition like cars or cats. Do you see it evolved to create mood categories like happy, sad, lonely ?
Yes.
With 880 billion images to be taken in 2014 according to Yahoo, isn’t sorting quickly becoming the number one issue to solve ?
Yep, that’s why we founded the company.
Could Euvision become one day the backbone of visual search online to compete with Google image Search ?
We’d rather help Google further improve their services than compete with them.
What would you like to add to Impala today that can’t be done because of technology restrictions?
Our aim is to make Impala faster and even more accurate. The limitation today is compute power on the target device. So we need more power on smartphones. With abundant power there, we will learn the smartphone to see anything.
You can download Impala from Google Play or Itune store for free.
Author: Paul Melcher
Paul Melcher is a highly influential and visionary leader in visual tech, with 20+ years of experience in licensing, tech innovation, and entrepreneurship. He is the Managing Director of MelcherSystem and has held executive roles at Corbis, Stipple, and more. Melcher received a Digital Media Licensing Association Award and is a board member of Plus Coalition, Clippn, and Anthology, and has been named among the “100 most influential individuals in American photography”
Paul,
Your comments are always perceptive. I enjoy reading them. But I wish you would segregate articles by relevancy to either 1) the Commercial Photography marketplace (i.e., the photograph itself as content); 2) Branded Photography (i.e., what you particularly enjoy: the photographer himself as content); 3) or the business of fulfilling “Kodak moments” for consumers. When all three are conflated as news about photography in general, it is confusing.
For instance, you wrote, ” . . .the 880 billion photos we will take in 2014 are nothing compare[d] to what is to come. But who organises all these images?” Well, I would respond by asking, Who cares?
Okay, ordinary CONSUMERS care. But it’s only the cloud equivalent (of the old paradigm) of stuffing drugstore prints and Polaroids in a shoebox. Sure, it would be nice to find that misplaced picture of Aunt Mildred at your cousin’s wedding. Still. it’s no more responsible to worry about how to sort such pictures than it is to preserve every off-the-cuff verbal remark one makes—in fact everyone makes, including every faux pas.
Taking snapshots has become just as ubiquitous, and for no more, or less, the same reasons, as has texting augmented speech telephony. But one could never accurately say that texting is literature, or that all of those billions of snapshots are Photography (with a capital “P”) worth keeping, let alone categorizing.
Tom,
Thank you for your comment.
I am not going to enter a debate on who is, or is not, a photographer. This debate is pointless. Only people that feel threatened put up boundaries.The reality is that the boundaries are getting blurred, faster and faster. We are here to discuss opportunities in photography and tech, wherever they come from. The judgement will be made on success.