Sinclair: Intelligently generating video highlights
Sinclair is one of the largest and most diversified television broadcasting companies in the United States. It has multiple national networks, live local sports production, and stations affiliated with all the major networks. It owns, operates or provides services to 191 television stations in 89 markets. Sinclair’s content is delivered via multiple platforms, including over-the-air, multi-channel video program distributors, and digital platforms.
Since Sinclair generates hundreds of gigabytes of video assets every day, it wanted a solution to intelligently produce relevant content clips for its audience and make that content easily searchable. Sinclair partnered with Egen and Google Cloud Platform to build a machine learning solution that helps it better serve viewers.
The solution initially focused on automatically identifying a weather segment in news programs. We created this using Google Cloud Platform’s Machine Learning APIs such as the Video Intelligence API, Speech API, and Vision API along with custom TensorFlow models. We integrated the resulting solution into Sinclair’s video streaming system so weather highlights can be made available as soon as a news program finishes.
Sinclair then worked with Egen and Google Cloud Platform to expand the solution for meeting other audience needs. For example, it can also identify the start and end times of specific segments within an hour-long news program, allowing viewers to find the specific content they are looking for quickly. Sinclair used the solution to further boost searchability of its video content for viewers by adding metadata to clips including the names of people and places, so the audience can easily use keywords to search for what they want to watch.
With this solution now in place, Sinclair can identify clips and generate highlight reels for a wide range of content, from news broadcasts to sports games. Watch the video below to see how Egen and Google Cloud Platform brought the power of machine learning to creating game highlights for a tennis match.
◼