Stock photo search engine Everypixel is a tool that should make the quest for perfect stock photos easier. But what’s even more interesting is their tool called Everypixel Aesthetics. It uses neural networks to tell you how “awesome” your photo is.
According to the developers, this tool sees the beauty of stock photos in the same way as humans do. So before you buy a stock image or upload one of your own, you can run it through this quick test and see what neural network has to say about it. I tested it out, and the results were surprising, to say the least.
The mission of Everypixel is to help designers discover awesome photos. Their tool is supposed to perform two important tasks. First, it will estimate a visual quality of every image and apply the aesthetic score to every file, which will enable automatic image curation. And second, the neural network would detect photos with the lowest aesthetic score and weed them out from the search results.
The idea of neural networks judging the beauty of photos intrigued me, so I played around with the tool a little bit. This is the first photo I submitted, one I really love:
Pretty good score for the first shot, right? When you click on the word “awesome,” you’ll get an explanation that the “service doesn’t measure the coolness or beauty of a person or any object in a photo. It cares only about technical parts like brightness, contrast, noise and so on.” It seems fair, considering that this photo is indeed technically correct. But then I went on playing. I submitted a snapshot taken during one night out. It’s grainy, blurred in places and pretty horrible when it comes to colors. Not to mention composition. But here are the results:
After this snapshot from a really fun night, I uploaded one of my most downloaded photos from Shutterstock. It scored less than the night out snapshot:
Finally, I remembered seeing someone posting a comment with a black brushstroke on a white canvas, with a score around 60%. So I gave it a shot myself. I drew this gorgeous work of art in Photoshop, and here’s the score:
This final result really made me chuckle. I suppose neural network is not really as accurate as the developers would like it to be, even in terms of defining the technical quality of the images. Still, I believe that the idea is good and it might become more developed and advanced in the future.
Although the artificial intelligence doesn’t really give the perfect results for judging technical quality of images, there’s one thing it does pretty accurately – tagging. All of the images got a set of suggested tags, most of which could be useful when you’re uploading photos to stock websites (or anywhere else). I always struggle to come up with enough suitable tags, so this can be a good tool for coming up with more of them.
And last, I must admit this is a great procrastination tool. It gives you a couple more minutes of fun before you force yourself to finally sit down and start editing that bunch of photos you took over the weekend. Try it out and tell us in the comments – how was your score?
[via DPReview]
FIND THIS INTERESTING? SHARE IT WITH YOUR FRIENDS!