Have you ever seen someone wearing something you wanted to buy but you didn’t know where to get it? Or maybe you saw something you liked while reading your favorite blog or flipping through Pinterest but you didn't know where you could buy it, let alone how to describe it in words? There is a saying that a picture is worth a thousand words, but who really wants to write that much when you are trying to find something quickly.
Today, eBay announced Find It On eBay and Image Search, two features that will make the entire internet shoppable. These new features bring you a step closer to getting that item you really want, whether it’s brand new or nearly new—it’ll be totally you.
Find It On eBay is a new feature in our eBay app and mobile platform that lets you share images from any social platform or web browser. All you have to do is “share” the image with eBay and our mobile app will find listings of the item in that image or others like it.
With Image Search, you can take a photo of something you want to buy—or use an existing photo from your camera roll—and put it into the eBay Search bar on our native apps. Then, we’ll show you listings that match the item you are looking for.
These features sift through the more than 1.1 billion listings on eBay, creating a seamless shopping experience and helping you find your version of perfect. They also open up new ways to discover unique and fun items that wouldn’t be possible with just using words.
Leveraging the latest advances in two core parts of artificial intelligence -- computer vision and deep learning -- these new features make it easier to buy the things that inspire you. When you upload images to run Find It On eBay and Image Search, we use a deep learning model called a convolutional neural network to process the images. The output of the model gives us a representation of your image that we can use to compare to the images of the live listings on eBay. Then, we rank the items based on visual similarity and use our open-source Kubernetes platform to quickly bring these results to you, wherever you are in the world.
We developed the idea for Image Search with a small team during eBay Hack Week, an annual company-wide competition challenging our technologists to innovate and reimagine the ecommerce experience. Our project won the competition in 2015, and from then on, our team has grown substantially while we’ve been building these features. We are continuing to work on additional features and expect to launch more computer vision products in the coming months.
Find It On eBay and Image Search will be rolling out this Fall. At the time of launch, Image Search will be supported on both Android and iOS and Find It On eBay on Android.
So, the next time you come across something you love when you are browsing online, snap a picture and find it on eBay.
Steve Neola is a Product Lead for Recommendations and Image Search at ebay. He is captivated by enabling everyone to use images to shop the world’s most diverse inventory no matter where they are or what inspires them. Steve is a huge music fan and can be found at jazz clubs in the West Village when not working at eBay NYC.
Ben Klein is an Applied Researcher at eBay, where he works on computer vision, machine learning, and deep learning. Ben leads eBay’s efforts on applying computer vision for search and recommendation systems applications. Prior to joining eBay in 2014, Ben worked at Microsoft Research as a machine learning researcher and developed algorithms that are used by Microsoft Xbox. Ben holds a Master’s degree in Computer Science from Tel-Aviv University and his work has been published in CVPR and ECCV.
Max Manco is a Lead Engineer for Image Search and is responsible for its design, performance and scalability. Max joined the eBay Israeli Office in 2011 and has been involved in several key projects in the Structured Data group. In 2015, shortly after relocating to NYC, he started working on Image Search. Max is very passionate about Java, the Spring Framework, and building efficient large scale systems that delight our users.