Wildberries has improved its image-based product search feature by updating the underlying neural network. The new system now considers the texture and small details of products to deliver more accurate results.
This update stems from training the neural network using millions of popular products from the marketplace. Wildberries claims that the accuracy of the search tool is now three times better than the previous version. The company initially introduced photo-based search in 2019, though the original neural network had been trained on a much smaller dataset.
With this new feature, users can upload an image, and the system will identify the specific product model or offer similar alternatives. The mobile app also allows users to highlight the area of interest in the photo, making searches even more precise.
Additionally, Wildberries recently launched a subscription service called “WB Club,” which offers access to a special section with exclusive discounts. The company’s founder, Tatyana Kim (Bakalchuk), has also hinted at upcoming services that will be introduced soon.
Source: vc.ru