Creating a video search engine creates a huge challenge in terms of indexing the content and tagging it with relevant keywords. VideoSurf solves this problem by applying Computer Vision to actually ‘see’ people and objects inside video.
Most video search solutions work by asking users to tag videos with information and extracting keywords from the titles to determine how they are relevant. Computer Vision actually allows the search engine to see people, places and objects within the videos to tag them more appropriately.
In addition to Computer Vision, VideoSurf makes it easier to find new videos by making it an altogether more visual experience. By applying the principles of visual search, navigation and discovery, the search engine makes it easier to find the video content.
VideoSurf’s visual search technology allows you to see much more than just a thumbnail of the video. By using scene detection to determine key frames, users can see multiple thumbnails from the same video to help find what they are looking for.
The search engine includes better navigation by allowing users to jump to different parts of a video. Clicking on one of the 10 or so automatically generated thumbnails eliminates the need to manually navigate through the video itself.
VideoSurf also helps you to discover more content by including related content in your search results. If you search for an actor, you may also get some results for other actors that they are connected to or have worked with. Or VideoSurf may display other actors in the same genre to help users discover new videos and actors.
This is the first search engine to successfully incorporate computer vision. Applying this sort of computing power to the number of videos on the web is a major challenge. If VideoSurf can apply these principles at scale, it could spell a much better way to find relevant video on the Internet.