Semantic technology and machine learning have many possibilities as the technologies continue to grow and evolve. This interesting topic came to us from Blasting News in their article, “Flickr presents visual ‘similarities’ in search results.”
The online photo sharing app, Flickr, has been experimenting with many new features over the years and the most recent involves an innovation to help their users find the right images in search results.
The new search feature uses advanced machine learning technology to come up with visually similar photos. The machine learning work behind the mapping of the raw pixels uses their current neural networks, which were trained to map pixels into a set of tags. The team added an extra step to plot the feature vector of an image and compare it with that of other images. Images whose feature vectors are located within the same visual neighborhood will be considered similar.
Defining the similarities was a challenge as there are many ways an image may be similar to other images and still remain different enough to not be identical. However this additional tool adds a new perspective to search beyond queries.
Melody K. Smith
Sponsored by Access Innovations, the world leader in taxonomies, metadata, and semantic enrichment to make your content findable.