Valossa, a video recognition and intelligence company, announced a major suite of and video insight tools.
Valossa, a video recognition and intelligence company, announced a major suite of upgrades to its video recognition engine and video insight tools. Notable improvements include the ability to analyze video streams for faces, objects and logos in real-time, as well as a renewed explicit content detector for recognizing levels of sexuality and violence. Valossa demonstrated the features at the recent National Association of Broadcasters (NAB) convention in Las Vegas last week, from April 9-12.
The company aims to help major media providers quickly utilize media resources that were previously difficult to discover and handle, and to provide appropriately targeted content for consumers.
“We are experiencing strong growth in participation to the Video Insight Tools preview program,” said Valossa CEO Mika Rautiainen. “Our video recognition engine is constantly improving in accuracy and breadth of recognition, while the scene-level searching and inspection with the Insight Tools now deliver valuable new insights from unstructured video data.
“The Tag and Train feature in particular is designed to help customers build their own library of recognizable personas. It is now very simple to add new faces to the face training gallery, where they immediately become part of the customer’s recognition arsenal.
“The most significant recognition improvement we are rolling out today is the analysis of content appropriateness for compliance use cases. Our recognition engine detects explicit, sexual, and sensual content both from audio and visual data, making it easier for customers and brands to find the best target use for their content. Real-time recognition of logos and famous personas will also bring new monetization opportunities for live video and broadcasts,” he concluded.
In addition, Valossa’s Video Insight Tools will be available as APIs that can be integrated to customer platforms, giving them the flexibility to utilize Valossa’s AI-powered content management capabilities in the familiar surroundings of their own in-house tools.
The full range of platform updates include:
Real-time stream content monitoring
- Video stream analysis for faces, objects and logos with our real-time recognition engine
Updates to Valossa video recognition
- Improved face detection and identity tracking (available in April)
- New explicit content detection for recognizing levels of sexuality and violence
- An updated object and scene recognition model with hundreds of new concepts and improved existing visual recognitions with millions of additional annotations
- Recognition of media structure elements like test patterns and clapperboards
New functionalities for the online Video Insight Tools Preview program
- Tag and train - the ability to build custom face recognition libraries directly from videos using online tools and via Face Training API
- Refinements in Valossa reports - better video recognition result validation through overviews and heatmap
- Video content searching at scene level based on audio-visual recognition data
- Custom keywords - highlight mentions and appearances of objects and people using custom word lists
Valossa launched the industry’s first commercially available video recognition and insights platform during the NAB Show in 2017, with the company currently seeing 50% growth rate in its user base quarter-on-quarter. The platform’s core advantage has always been the focus on multi-modal recognition, delivering comprehensive audiovisual recognition and content analytics via simple APIs and tools.
Product Adopted:Digital Video Recording