Pattern Recognition Shopping

The new Modista online store collects samples of goods from hundreds of retailers and stores them in a single database (163,000 products in four categories: shoes, watches, handbags and glasses). Then, on this base, an image recognition engine with self-learning elements is launched.

Search for purchases is carried out exclusively through the visual interface. Click on the product you like most - and the table is rebuilt for the new template. Horizontal - similarity in shape, vertical - similarity in color.

It can be assumed that similar interfaces in the future will become a standard element of any online store.



Of course, this is not the first project of its kind. In recent years, several similar product search engines have been launched. The most famous of them - Like.com, which allows you to filter visual search results by manually selecting a style, color, brand, material, and even selecting sites of specific online stores from the list. Moreover, you can manually specify the fragment on which you want to set the pattern recognition engine on the product photos. The screenshot shows how to select a fragment.



Or here's another interesting Browse Goods project . This is a visual classifier of all products with a zoom in the style of Google Maps. True, there is no pattern recognition here, but this does not detract from its aesthetic value.



Here are links to other visual interfaces for shopping.

Amazon Windowshop
Zoomii Visual Amazon Store
Amazon Book Map
Map Amazon
AmazNode
Oskope Amazon Information Graphs
Viewzi Book View
Vizzl shopping engine
BlackDogAir
Music Plasma (music search)

If some projects have ceased to exist, then the description and screenshots can be found here .