Testing a neural network, which helps to find beautiful photos on image websites
To teach the search engine to automatically distinguish fashionable photos from outdated ones, this is the task which the developers of Everypixel faced. To find out how well this approach works, we compared the search process on Everypixel with the popular free image websites Pixabay and Freepik.
At first glance, all free image websites are the same. But, as soon as it comes to searching, the difference becomes apparent. And, it’s not even in the content, though it’s important. The difference is in how well the search engine works.
Traditional search engines can’t see images, as humans do. They show pictures for your search query based solely on the description of photos. However, the description refers to the content and does not say anything about the style and quality of the picture. Therefore, the search for a suitable style of photography on the stock photo sites becomes a challenge.
Everypixel is the first search engine for stock photos, which uses artificial intelligence to curate content.
To help users quickly find beautiful photos, the developers of Everypixel decided to make the search algorithms smarter. They created a neural network and trained it to distinguish trendy photos from dated stock images using the dataset of 956,000 photographs. You can learn more about Everypixel AI and even ask what it thinks about your pictures here.
Now you can decide whether the developers have coped well with the task at hand, or not.
Experiment
To show how the neural network works in practice, we compared the Everypixel search results with the search results on free image sites Pixabay and Freepik. These websites have many photos in different styles, and there are plenty to choose from.
We decided to start a comparison with a business topic, which is a very popular category in all stock photo sites. Compare the result:
Denying the established cult of coffee in modern society is difficult. Besides, freelancers often work from a cafe, so photos of beautiful coffee houses are a favorite topic on image sites. Let’s see what results we will get with the ‘cafe’ request.
After business and coffee, we’re sure you’ll want to see something relaxing, so we made a request by the keyword ‘happy.’
As this experiment showed, the Everypixel neural network understands which photos are now trending and shows them at the top of the search results.
So, what do you think? Does the AI help image lovers to find contemporary stock photos faster? Was it helpful in your case? Share your opinion in comments.
You can try the benefits of an image search engine powered by AI on everypixel.com
