I have a box of old photos in the corner of my office that needs sorting. It contain about 500 photos, all of which need scanning and backing-up, or discarding. Will I ever complete that task? Probably not. The same goes for the three SD cards on the shelf and the ‘photos to sort’ folder on my desktop that’s nothing more than a dumping ground after each holiday or trip.
The internet has a more pressing problem that will only get worse if I do ever get around to ‘clouding’ my photos. A staggering 880 billion photos were uploaded in 2014. Instagram alone gets 40 million every day, while Flickr’s collection now sits at around six billion. I’m guilty of polluting both.
How to automatically sift through the almost uncountable images uploaded to Internet is one of the biggest technical problems of our age. The reason? We’ve long passed the point where we can see them. Most photos on the internet – especially those uploaded en masse after the kind of stock-take on my to-do list – are digital clutter, no more likely to be viewed again than the box in the corner of my office. However, there are now ways of autonomously sifting and sorting.
Online photography community/marketplace EyeEm, based in Berlin, has developed an algorithm that’s capable of finding beauty – or not – in a photo purely on a pixel level, whether the photographer is a professional or not. The algorithm captures the patterns that are unique to great photographers, thanks in part to the code being built around feedback and judgements from a group of top Berlin photographers.
Software good enough to match the taste of a person instead of a machine is quite something, but what about identifying what’s in an image? That’s something being worked on by British computer scientist Stephen Wolfram, creator of the Wolfram Alpha computational search engine. He’s created ImageIdentify, which is available online; simply drag in a photo, or choose one from a phone’s camera roll, and it will identify the contents. The computer language is based upon symbolic pattern matching and can be applied to huge archives of photos, but make no mistake about it; this is artificial intelligence being put to practical use.
There’s no longer any need to drown in data, but there is still a problem; what to do with all of these photos? Researchers from the University of Washington recently turned the sheer volume of photos into an opportunity. They took 86 million timestamped and geotagged photos from Flickr, and identified locations they had multiple photos for. They then warped each to a common viewpoint, and generated a short time-lapse video for each one. Cue photo-stiched time-lapse videos of a Croatian waterfall changing course over several years, a Norwegian glacier retreating, and a New York skyscraper slowly rising – and all from a bunch of holiday snaps.
Despite the big numbers, photography is less of a problem for the cloud than video. The number of smartphones and tablets that can film video is due to double to two billion worldwide by 2017, though it’s GoPro, the wearable and mountable HD camcorder, that’s getting much of the blame. GoPro reports that in the first quarter of 2015 the word ‘GoPro’ appeared in the title, file name, tag or description of around 6,000 YouTube uploads each day, and accounted for more than a billion views of 50 million hours of videos.
That’s a lot of boring bike rides. Thankfully, companies like Antix and RipIt are issuing apps for skiers and snowboarders, and surfers, respectively, that automatically edit GoPro videos before uploading. Two-hours of footage can be automatically cut down to a two-minute highlights reel (or, in the case of GoPro bike ride videos, delete them in their entirety). With 4k video now possible from GoPro, that kind of editing comes just in time to prevent the web being overun with high-resolution mediocrity.
Automated identification and editing might seem rather dull, but it helps to both reveal great stories and creates an online meritocracy. If you produce a great photo or video, the internet will find it, use it, and show it off. In a world where the total number of photos taken is roughly doubling each year and where video is going ballistic, that’s reassuring. Just don’t expect any feedback.