Sixth Sense Retrieval – Making Sense of News Video Archives in Seconds

Web video content is in high demand these days. As a consequence, most news organizations and media companies have built impressive archives. One big challenge remains, though: It’s often hard to find the video content you’re looking for. Using the latest in artificial intelligence (AI) technology, Sixth Sense Retrieval offers a solution here.

Sixth Sense Retrieval is a joint venture of Latvian startup Mindflux and our team (= DW Research & Cooperation Projects), partly funded by Google's Digital News Initiative (DNI). Its ultimate goal is to provide an image-recognition solution that's not only effective, but also easy to implement. The prototype, under development since February, already offers a set of very useful tools like image segmentation, object detection, and scene summarization. Sixth Sense Retrieval analyzes video content and automatically provides annotations, thus speeding up production workflows.

Using this system, journalists and editors will be able to find the right video within seconds–even if they're not particularly IT-savvy, and the clip has been stashed without metadata, in a remote corner of a vast, poorly structured video archive. Sixth Sense Retrieval makes sure media producers can spend more time and resources on creating high-quality content, as they don't have to worry about cumbersome tagging and searching anymore. The platform does the heavy lifting.

Sixth Sense Retrieval will run until the end of July 2018 (with a follow-up project in the making) and complement our projects SUMMA and news.bridge–also AI-driven, but with a different focus.

If you're interested in testing Sixth Sense Retrieval or would like to find out more about computer vision and video retrieval, please find us on Twitter.

Alexander Plaum
Peggy van der Kreeft
Olga Kisselmann