Special Issue _____ Development of Broadcast Platform Tailored for Viewers
Share
Magical Technology to Rewind Time in Sports Games
Play Favorite Moments of Sport Events Again
In general, viewers watching sports games on television often miss important moments or want to watch specific scenes again. However, viewers do not have the discretion to select and watch the broadcast images. They simply have to repetitively watch images they do not like, or wait until the scenes they are interested in appear. The videos uploaded to major portal sites after each game ends usually contain the highlights, leaving many viewers unable to find the scenes they want.
A group of ETRI researchers has successfully developed broadcasting platform technology to automatically create and replay videos containing major events in sports games such as basketball and baseball, based on sports broadcasting images and text message alerts. Called “broadcasting platform technology tailored for viewers,” this technology magically rewinds time to past scenes. Users can select and rewatch specific event types such as basketball shots or home runs in sports broadcasting images for basketball and baseball, and they can also watch highlights for a specific team or player of their choice.
Customized Highlight Videos for Better Viewer Experience
The ETRI researchers automated the service to provide the videos in real time. As a result, videos with a quality equivalent to manually edited videos are provided within tens of seconds. Since viewers can configure the events of their favorite players or teams, the videos provided are specifically tailored as highlight videos for each viewer.
The broadcasting platform technology tailored for viewers obtains from text messages alert information on various events such as shots, rebounds, hits, and home runs in actual sports games. It then recognizes the screen layout or players’ movements in the broadcast videos to segregate the relevant event sections. To this end, the computer directly classifies and extracts the scenes of pitching as well as infield and outfield, and information about the game time and number of pitches through deep-learning technology.
The image information extracted in this manner was analyzed for the patterns used for creating broadcast images, at which point machine-learning algorithms are used to identify event sections. Lastly, the computer matches the events identified from text message alerts to the respective sections in broadcast videos, thereby creating a final video with information about event sections.
“Watch Again” Service to Realize Fully Tailored Broadcasting
The viewer-tailored broadcasting platform technology is expected to be embedded in the services that will automatically classify sports game images in which viewers are interested from the real-time streaming broadcast or digital broadcast and provide them within tens of seconds for replay.
IPTV operators can apply this technology to their set-top boxes to offer customized replay services to their customers. In addition, when connected to mobile devices, the technology will be able to provide television viewers with replay services without interrupting their viewing experience. The technology will also enable archiving service providers to provide their customers with automatically indexed images in large volumes.
The ETRI researchers are planning to expand the technology’s application to more types of sports in addition to baseball and basketball that have certain patterns for image production, which will result in business partnerships with or technology transfer to IPTV operators, sports broadcast archiving service providers, and sports clubs that need strategic analysis.
In the meantime, the technology was introduced at NAB (National Association of Broadcasters), earning favorable reviews from various participants including FOX Sports, which broadcasts NBA games.
