Powered by Traco Systems, Traco AI is a comprehensive media processing solution based on Face recognition and Object detection. Platform has been developed for advanced metadata enrichment with help from automatic speech transcription. 

This solution has been developed with the intention of simplifying some lengthy TV production processes. The result is a system that can save hundreds of hours of media content analysis.

Recently, this functionality has become a part of Octopus X, a new version of MOS based newsroom computer system designed for use by television and radio broadcasters.

How does it work? 

Traco AI uses machine learning technology and neural networks. Recognition process is based on detailed biometric mask working with the principle of nasal and ocular vectors. Therefore, detection accuracy of visible faces works perfectly. This platform can act as a standalone server within an internal domain which is its advantage over other cloud solutions.

What are the main features of Traco AI?

  • automatically recognizes faces based on biometric identification
  • allows to search for groups of people when they are in one shot
  • enables to create a database using photos, archive footage or online databases
  • can be integrated into any Media Asset Management (Avid Media Central, Viz One, etc.)
  • offers automatic subtitle completion in broadcast, based on the internal database
  • provides speech-to-text recognition and automatic closed captioning generation (CC).

There is a wide range of ways Traco AI can be used:

In news production, it can increase effectivity by identifying faces that frequently appear on air. After simply entering keywords, the user will see a list of all the scenes where the person appears. 

Sport content producers can save time by letting Traco AI analyze the obtained video material and mark all the visible faces of the athletes. Producers can simply add them into their database together with some basic metadata. Apart from that, it can also recognize the score changes. Afterwards, the editor can initiate ongoing processes in a simple user interface to streamline the production workflow.

In reality show production, Traco AI can help by making collages from all the recordings which illustrates each finalist’s journey through the competition. By creating a database of participants, editors just enter a name and see a list of scenes where the contestant appears.

While managing an archive with countless hours of shots taken over several decades, AI Facer can recognize whether it is the same person in different stage of life. 

We developed the Traco AI platform as a Full Rest API, which allows integration to various workplaces. Every production can decide what kind of functions they find useful for their projects or which type of interface they like the most. 

If you want to find out more about this platform, visit ai.tracosys.eu. Demo of this platform will be showcased at IBC 2019 (7.A39).