Similarity Search Comes to Watson Visual Recognition

Today, we are excited to introduce Similarity Search BETA to Watson Visual Recognition, allowing users to find images similar to a queried image. The new feature utilizes machine learning technology to recognize visual content and organizes collections of images accordingly.

Users can upload their own collection of images and the service will then return the top 100 most similar images. Similarity search can be utilized for any purpose, and users can custom train the service on collections of up to 1 million images each.

Similarity Search Tackles Retail

Our customers are already using Visual Recognition similarity search to enhance retail experiences. Here are a few examples we’re excited to share!

The Hunt Cures Outfit Envy with Cognitive Computing

The Hunt team selected the Visual Recognition capability from a dozen competing image recognition solutions because of the power of its cognitive abilities, which not only match images, but evaluate them, regardless of image distortion.

Yoox explores the capabilities of Visual Recognition Similarity Search

Italian internet mail order retailer of men’s and women’s multibrand clothing and accessories has tested Watson Visual Recognition to improve the “find similar” customer experience by identifying what a client is wearing when they enter a store, and then determining what they would most likely want to buy. The customer can also speak with the site’s chat-bot to complete a similar image search.

Get started now by clicking on the button below!

Try the new similarity search feature