As video clips continue to proliferate online, companies such as VideoSurf and Digitalsmiths have developed search tools to help web users sift through all the choices, reports the New York Times. Traditional engines that search video typically rely on keywords, descriptive captions, or other tags, often supplied by the producer, to identify the contents. Engines that go beyond these descriptive tags to search the visual contents themselves allow researchers to get under the hood to see what’s inside the video, instead of counting on tags that may be limited or even inaccurate. With these new tools, users can even locate an exact segment they want to view without having to click "play" and watch the entire video. VideoSurf, based in San Mateo, Calif., has a free video search engine that is now in beta version at It uses computer-based facial recognition and other tools to analyze videos frame by frame and to identify characters, objects, and scenes shown within each frame for detailed searching. Type in "bailout," for instance, and many of the characters in this national drama appear in thumbnail portraits across the top of the screen, above the general search results. Click on "Henry Paulson," and the results are re-sorted to show videos including him. Digitalsmiths, based in Research Triangle Park, N.C., also uses facial recognition and other computer vision techniques to index web videos. Unlike VideoSurf, which has a direct portal for consumers, Digitalsmiths puts its search engine to use behind the scenes, within the web sites of companies that want to make their videos more accessible…

Click here for the full story

About the Author:

eSchool News

Add your opinion to the discussion.