Picture this: introducing the Descriptive Camera

May 16th, 2012 by

microtask_descriptive_cameraPhotography has come a long way. Two centuries ago taking a family photo meant standing perfectly still for ten minutes in a pose stiffer than a waxed mustache. Now we’ve all got multi-megapixel cameras built into our phones (or cans of processed meat), and if we want that scratchy sepia look without the waiting around, we can just use an Instagram filter.

This is great, especially for people like me who like to take photos. The downside is that the more photos we have, the harder it is to sort them in a way that makes them useful in the future. The trigger happy amongst us now feel like we are drowning in a sea of digital photos.

Before you cast away your digital cameras and save yourselves (yes, I’m exaggerating the problem slightly), hope is here. It’s called the Descriptive Camera, and it intends to bring a whole new perspective to photography.

Say ‘Crowd!’

The brainchild of ‘creative technologist’ Matt Richardson, the prototype Descriptive Camera does exactly what the name suggests. You point the camera and press the shutter button, but instead of a photograph, the built-in printer spits out a short text description of your image (it turns out a picture is in fact worth about 150 words).

So how does it work? Advanced pattern-recognition software? Quantum Entanglement? A tiny prehistoric bird? The answer, as regular readers of this blog will probably have guessed, is crowdsourcing. The Descriptive Camera uploads images to Mechanical Turk, where users are paid a small fee for providing a brief text description. Thanks to the popularity of Mechanical Turk, and the relatively high price per task of $1.25, Richardson says that it typically takes between three and six minutes to ‘develop’ each photo.

Tell me a picture

The Descriptive Camera was designed primarily to help categorize all those photos on our hard drives. By including the unique description in each photo file, the Descriptive Camera allows text-searching of photographs. Looking for a picture of a mountain? Just search for mountain. A sunset? Search for sunset. Sounds great so far, but there are limits: how about a picture of your parents? Anonymous Turkers are unlikely to be able to recognize your mother and father, or be familiar with your friends and pets (at least we hope they won’t). Despite this potential limitation, the Descriptive Camera, (which is still in the experimental stage), may prove an important step on the road to more manageable images.

But as well as offering a potential solution to a cluttered ‘My Pictures’ folder, the Descriptive Camera is also a great example of what can be done with crowdsourcing platforms like Mechanical Turk (and say …ummm… Microtask, for example). It’s now possible for anyone to bring the crowd into their project, whether it’s a towering data processing task or a quirky prototype like the Descriptive Camera.

This means that innovative thinkers like Matt Richardson are free to pursue unorthodox ideas and create unique crowd-based projects. But it also shows just how easy it’s becoming to incorporate the crowd into everyday life. As the Descriptive Camera shows, it’s entirely possible to fit a whole crowd in your pocket.