The New Video Analysis Intelligence Update is Now Live

March 6th, 2017
The video analysis API got the second intelligence update and we invite you to test the improved performance level and new useful features:

The upgraded metadata format

The Valossa Core Metadata has been redesigned to adapt to your video analysis needs better. The format upgrade includes a cleaner, unified and extensible logic that allows you to integrate it easily to different applications:

  • Insights for every second of the video
    All detections will now be indexed for every second of the video. You can browse inside the content details and find who and what appears at different time points, grouped by each second.
  • Detection groupings refine your results
    The detection groupings describe the most dominant concepts that represent a particular segment of a video. You can see for example in which La La Land scenes John Legend appears or when the characters were talking about jazz.
  • Filter results by confidence level
    The visual context and face detections include the confidence level information that can be used to filter the analysis results. This allows you to decide the level of accuracy that is appropriate for your application.

Video categorisation and visual context

The video AI can now recognize visual context, such as objects, places, styles and activities, with higher confidence and detail than before. Your content will be also categorized according to the standardized IAB taxonomy, helping you to easily classify video inventories.

API for image tagging in closed beta

A brand new Valossa Image Tagger API is in closed beta. If you would like to test the tagger with your images, contact us at and we’ll help you to get started.

Latest posts