Schibsted Media Group's Austrian market place Willhaben has integrated its new feature “Fashion-Cam” in its app, an automated image recognition tool that is based on Artificial Intelligence and Deep Learning concepts. As complex as this technology may be – willhaben users will find it incredibly easy to work with this new feature.
“Willhaben users who are passionate about fashion can use this new feature to take pictures of their favourite dresses or other garments, whatever and wherever they like (such as clothes items in magazines, out on the street), and willhaben will automatically look for suitable fashion items on our platform that match in terms of garment colour, pattern or shape. This makes browsing the world of willhaben fashion even more fun, much easier and in most cases also considerably faster. No matter where and when our users encounter an intriguing piece of clothing, they just have to take a picture and willhaben will do the fashion hunt for them,“ Sylvia Dellantonio, General Manager of willhaben explains.
In contrast to a traditional search approach using search items and filters, the Fashion-Cam operates on the latest Machine Learning technology: using Artificial Intelligence, pictures are analysed based on their visual components such as colour, pattern or shape and this data is then compared to the willhaben-content to find a match. In addition, the feature has also been generally extended within willhaben – while browsing fashion ads, any of the pictures can be used to perform an immediate search on similar items.
”In the last couple of months, our system has been perpetually trained to operate on more than two millions of images and is continuing to automatically learn by every image it encounters. Already, the results yielding from this unique search are very good, providing the visually most similar products on willhaben,” Dellantonio sums up.
The Fashion-Cam feature was developed by willhaben in close cooperation with the Styria Digital Media Group data science team.