Visual Search Classifier block: Classify images by visual similarity

A new Visual Search Classifier block is available in Workflows. It classifies an image by running a visual similarity search against a labeled dataset and returning the label of the closest match, so you can get classification results without training a dedicated classifier.

This helps when you have a large number of classes, where training a traditional classifier is impractical, for example identifying a specific product in a large catalog. The block builds on the Roboflow Visual Search block and returns a standard classification result, so you can drop it into an existing Workflow and route its output to downstream blocks.