Facebook’s artificial intelligence (AI) team has built a visual search system that can recognize content that appears in photos and return relevant search results.
Called Lumos, Facebook originally created the platform so that its visually impaired users could understand the content of photos. But Facebook recognized that everyone could benefit from this type of visual search system.
Facebook’s image search system can detect and segment objects, scenes, animals, places, and clothes that appear in images or videos – and understand them.
For instance, let’s say you search for “black shirt photo.” Facebook said that Lumos will search for and see any photos that contain a black shirt – even if no tags have been added to the photo. Facebook will then return search results that are relevant to the query, as well as diverse.
“We’ve built a search system that leverages image understanding to sort through this vast amount of information and surface the most relevant photos quickly and easily,” according to a Facebook blog post published today. “Using Facebook’s automatic image classifiers … you can imagine a scenario where someone could search through all the photos that his or her friends have shared to look for a particular one based on the image content instead of relying on tags or surrounding text.”
Facebook said its AI team used “cutting-edge deep learning techniques to process billions of photos and understand their semantic meaning.” Lumos is continually improving through machine learning, according to Facebook.
You can read more about this in Facebook’s blog post.
Image Credit: Facebook
Subscribe to SEJ
Get our weekly newsletter from SEJ's Founder Loren Baker about the latest news in the industry!