Washington: Have your ever felt riled at the irrelevant and unwanted results a search engine throws in response to your query?
Don't worry. A new computer programme developed by scientists - including an Indian-origin researcher - at University of Washington and Allen Institute for Artificial Intelligence in Seattle is adept at teaching itself everything available on internet about a comprehensive visual concept.
Called “Learning Everything about Anything” (LEVAN), the programme searches millions of books and images on the web to learn all possible variations of a concept, then displays the results to users as a comprehensive, browsable list of images - helping them explore and understand topics quickly in great detail.
Major information resources such as dictionaries and encyclopedias have limited coverage as they are often manually curated.
“The new programme needs no human supervision and, thus, can automatically learn the visual knowledge for any concept,” said Santosh Divvala from Allen Institute for Artificial Intelligence.
The programme learns which terms are relevant by looking at the content of the images found on the web and identifying characteristic patterns across them using object recognition algorithms.
“It is all about discovering associations between textual and visual data,” said Ali Farhadi, an assistant professor of computer science and engineering at University of Washington.
“The programme learns to tightly couple rich sets of phrases with pixels in images. This means that it can recognise instances of specific concepts when it sees them,” Farhadi explained.
The research team will present a related paper this month at the annual conference titled “Computer Vision and Pattern Recognition” in Columbus, Ohio.