Recommendation SystemVerified

Unsplash deployed Visual Search, a machine-learning-powered feature that lets any user upload or paste an image to find visually similar photos in the Unsplash library. It launched in January 2021 and is available to all users at no charge.

Details

Visual Search accepts an uploaded image or URL as input and returns ranked results of visually similar photos from the Unsplash library, sorted by relevance according to a machine learning algorithm. Unsplash describes the system as attempting to understand what is visually relevant in the sample image. The feature is accessible via the search bar on the Unsplash website for all users.

Products affected

Unsplash (web platform)

Sources & Evidence

Company Disclosure

Introducing Visual Search

Unsplash Blog·Jan 2021

https://unsplash.com/blog/introducing-visual-search/

Company Disclosure

Visual Search | Unsplash Help Center

Unsplash Help Center·Jan 2021

https://help.unsplash.com/en/articles/4680447-visual-search

Other practices by Unsplash

ModerationUnsplash uses a content filtering system built into its API that automatically screens search results to prevent images violating submission guidelines—such as nudity or violence—from being returned to API consumers. This is active by default for all API applications.RecommendationUnsplash uses advanced image recognition technology to power keyword-based search across its library, with the Unsplash API officially described as being powered by 'tens of millions of community generated tags and advanced image recognition.' This is available to all users and API consumers globally.Data AnalysisUnsplash deployed a machine learning-based visual matching system that scans the public web to identify where contributor photos have been used, then surfaces that information to photographers in a 'Uses' dashboard. This feature launched in May 2021.

Have evidence about Unsplash's AI practices? Submit a report.

Report a Sighting →