Amazon's New AI That Recommends Clothes

Have you ever really liked a shirt or a dress but you failed to put its greatness into words so you can't find something like it online? Well, Amazon's new StyleSnap AI is the answer to all your troubles.

StyleSnap incorporates computer vision and deep learning into part of the Amazon mobile app to find similar clothing items from a photo. It has been compared to Shazam, which finds the name of a song if you play it to the app.

In a blog post by Amazon, they explain the intricacies that go into making an accurate algorithm that can look at a photo of a skirt and then find something similar. They explain that they use neural networks made up of several layers where the first few layers are for the basic concepts such as the general shape and colour of the item and the middle layers identify patterns like polka dots. The AI can then give the user a selection of clothes that are similar to the photo and are available on Amazon.

They also went on to say some of the challenges they needed to overcome while perfecting the algorithm.

For example, if we feed a network thousands of images of maxi and accordion skirts, it will eventually be able to tell the difference between the two styles. If we, however, present it with one Scottish kilt, it may be confused and predict an incorrect class until enough examples are provided to train it otherwise.
— Amazon

Furthermore, the company also said that fashion influencers who are part of the Amazon Influencer Program can get commissions from items they helped to sell.