In our increasingly visual world photos tell the story. Social networks have been early in recognizing this, together with the human urge to post, share, view, like and comment on photos. Images have become the driving force behind social network growth – helped by (auto) tagging and awesomizing features. On average 300.000 images make their way to Facebook alone every minute, on New Year’s Day and Eve this nearly doubles. A rough estimate brings that to over 300 billion photos on Facebook alone. Now that’s Big Data.
Big ‘Cold’ Data
To host all these picture Facebook are building ‘cold storage’ data centers for less popular or outdated pictures. Facebook is aiming to have the first of three facilities operational by fall. Each of the 16,000 square-foot data centers will be able to hold an Exabyte of data, or about one million PC hard drives. All these pictures are being indexed. And that won’t be all. Social networks will not only index the pictures, from their computer vision systems backend – machine learning systems that have the power and intelligence to identify what’s in an image, what a building looks like versus a face versus a landscape – they will know exactly what’s in these pictures.
It won’t stop here. More camera’s, phones, apps and social networks now have GPS enabled so users can display their photos by location with a mapping feature, utilizing the device’s geolocation abilities to visually showcase where photos were taken. Facebook, Instagram, Google+, Foursquare, to name just a few, have this all enabled. Also EXIF (Exchangeable Image File Format) data – small data files embedded in pictures – can show exact GPS location – revealing where photos have been taken and where people in these picture have been hanging out.
A Sourcer’s Goldmine
So if social networks know who and what is in these pictures and where they’ve been taken, sourcers should be able to find out as well. After all, since sourcing is about finding people, every online trace will give us more information about the whereabouts of our potential targets. And believe me, with the amount of pictures people have floating all over the internet, sometimes without even knowing of their existence, there’s enough traces to give sourcing a completely new dimension. Images have become a sourcer’s goldmine.
Avatars or profile pictures usually give fantastic results for visual sourcing, since they are mostly showing faces. Many people spending their time online have the habit of using one single picture for different online profiles, making it fairly easy to find them in the different social networks they use. Reverse image search in Google Image Search of a person’s profile picture will easily point to profiles with identical pictures.
Interestingly, one of the benefits of visual sourcing actually is that it can deliver more relevant results as opposed to just name searching. Especially for more common names, images prove to better identify a unique person.
But there’s more than just profile pictures. Think of pictures on company introduction pages, event pictures, pictures used for online check-ins on Foursquare, photo sharing websites like Flickr and much more.
Face Recognition Technology
Face recognition technology, fascinating technology, already heavily used for finding criminals, will change image sourcing fundamentally. People who frequently travel have had their faces being recognized by advanced computer systems while passing international boarders. Just imagine the data that governments collect with that. Face recognition technology is already being used commercially, to adapt product offerings in digital commercials for men or women passing high street stores, checking in using their online profiles or to accept payments.
Facial recognition technology can already identify people with 99 percent accuracy under the best circumstances. Best circumstances for facial recognition technology still mean having an ideal probe image and a database of similarly ideal images for comparison as possible matches, like headshot photos similar to those seen in passport photos or mug shots, however the power of machine learning systems and improvement of facial recognition techniques will quickly decrease dependability of best circumstances. Faces then will be recognized in all circumstances.
Facial-recognition algorithms can already be pared down and streamlined to run on the limited computing power of a smartphone or Google Glass devices. Did you also notice search engines and social networks having this obsession for facial images all of a sudden?
OK Glass, Source Now!
Oh yes, that’s where we could have been now. Having face recognition technology available in a high-tech device called Google Glass or other wearables would give us sourcing superpower, making true spies of every single sourcer. If it wasn’t for privacy concerns, this will become a major breakthrough in talent sourcing. Finding information of people based on the visual evidence they have available online through the sources mentioned above, triggered by the OK Glass, Source Now command. Sourcing will never be the same
This blogpost has previously been published on UK Sourcers as part of the Discover Sourcing Event organised by Katharine Robinson on September 17-18, 2013 in London. Below some of the video’s covering the content of the sessions.
Below you will find the full Discover Sourcing presentation about Visual Sourcing.