Insys Video Technologies Automates Celebrity Identification and Inappropriate Content Detection through API Integration with Amazon Rekognition
May 16, 2019 — Dübendorf, Switzerland: Insys Video Technologies has added powerful new capabilities to its comprehensive, over-the-top (OTT) video solutions that enable automated identification of celebrities and detection of inappropriate content in catch-up TV programming and on-demand video libraries. Viewers can find clips featuring a particular celebrity and jump to the exact point in the show where that person appears, while nudity and violence can be automatically flagged and blurred.
OTT video services enable media organizations to offer more content to their viewers than ever before. However, the huge volume of available content creates challenges for both audiences and providers. Consumers need efficient ways of discovering shows that interest them, which can be like finding the proverbial needle in a haystack.
Meanwhile, operators must ensure that all content meets their brand’s standards an regulatory restrictions for sensitive material, but the sheer number of clips makes manually reviewing them cost- and time-prohibitive. Recordings of linear channels for catch-up viewing are particularly challenging, as they typically do not include the same level of detailed metadata that accompanies “pre-packaged” on-demand programs.
Leveraging API integration with Amazon Rekognition, a solution from Amazon Web Services (AWS), the new features in Insys’ OTT platform can solve both problems. Content added to the operator’s library is quickly and automatically analyzed for recognizable people and sensitive content. Amazon Rekognition is a deep learning- powered video analysis service that can identify objects, people, text, scenes, and activities as well as detect any inappropriate content. The scalable service performs real-time video analysis using machine learning models that are trained to detect Insys Adds ML-Based Celebrity Identification and Inappropriate Content Detection to OTT Platform – 2 thousands of objects and activities accurately, and to extract motion-based context from video.
Insys’ rich front-end user experiences utilize the results of this analysis to allow viewers to search for shows or news clips that include particular celebrities or politicians. A timeline then allows them to jump directly to the specific moments at which those people are seen, while overlays in the Insys video player show the names of recognized individuals during playback. The Insys player window also displays markers on the program timeline to indicate scenes with nudity or violence, with the operator-controlled option to automatically blur the video during playback to obscure the offending content.
“Our new inappropriate content detection and blurring capabilities help operators automatically ensure content meets the expectations of their viewers and their brand,” said Insys CEO Krzysztof Bartkowski. “They also further expand our toolset for ensuring safe viewing for children, which we consider an important part of our corporate responsibility as a vendor. Meanwhile, the ability to identify celebrities and politicians within the video enhances content discoverability for service providers’ audiences, letting viewers spend less time finding relevant content and more time watching it.”
Insys Video Technologies offers a complete range of flexible OTT tools, technologies and white-label solutions that enable media enterprises to easily manage, monetize, distribute, and promote their content through branded, full-featured websites and apps.
InsysGO lets MVPDs and service providers launch comprehensive OTT offerings encompassing many channels in as little as 30 days, while Insys OneChannel delivers similar functionality with a streamlined viewer experience tailored for presenting a smaller number of channels. Insys VoD allows content providers to easily offer their video libraries for on-demand viewing, while the modular Insys OTT Toolkit enables customers to build their own customized OTT platform using modular components.