Kaemingk: Visual Search

Kaemingk is a prominent importer and exporter of seasonal decorative products. Each fall and spring, the company showcases its creative home decor collections for occasions like Christmas, Valentine’s Day, and Easter, as well as for the spring/summer season. Their inventory features around 20,000 home and garden decorative items, which are distributed to numerous professional clients across more than 80 countries globally.

Challenge  

Kaemingk collaborates with manufacturers in Southeast Asia to produce their items, often planning about a year and a half in advance for their seasonal collections. To introduce the most innovative and desirable products, they rely on a trend team that attends trade fairs for insights into upcoming trends. During these fairs, team members capture photos of various products, which serve as inspiration for new designs produced exclusively by Kaemingk’s manufacturers. However, the countless photos collected each year are organized in folders by trade fair without the addition of metadata such as colors, styles, or subjects due to the overwhelming volume, making this process impractical.

As a result, locating specific photos becomes a challenge, often requiring significant effort. There is currently no way to search these images based on criteria like subject, design, color, shape, or finish, which prolongs the design process and increases the time and labor involved in collaborating with manufacturers. Therefore, Kaemingk sought a solution to enhance the accessibility of their photos.

Solution  

Squadra: Machine Learning Company provides an advanced AI-driven photo search tool known as Visual Search. Utilizing vector technology, its algorithms can analyze the content of images without relying on textual (meta) data. This functionality allows users to search Kaemingk’s extensive photo collection via a dedicated search page by entering terms related to topics, shapes, colors, styles, and designs. For example, searching for a “white side table in Italian design” would yield the corresponding images in the search results. The search engine is available in English, making it accessible for international use.

Result  

With the implementation of Visual Search, Kaemingk has gained an efficient image search engine that dramatically reduces the time required to find relevant images from their vast collection within seconds. Besides significant time savings, the identification of trends is expedited, and the time needed for designing and supplying manufacturers has been considerably decreased. This innovative solution from Squadra Machine Learning Company gives Kaemingk a competitive edge in the market.