Our products is composed by different modules that identify different concepts which are then used to create the semantic analysis output.
Using more modules gives a deeper analysis but they can also be used as stand alone services to identify specific concepts
We're constantly refining our modules and taxonomies and developing new modules. New modules or existing ones can be trained to meet specific needs.
Knowing which talents and character appear in content is a basic information which can be related to other information extracted from video content.
Our Face recognition module detects when and where a face known by the system appears in a shot. Out of the box the module is trained to recognize 10.000 worldwide celebrities.
Custom tailored face recognition is also possible.
Thanks to our system architecture we can train neural nets to recognize new faces in a short time.
Sensitive content detection.
Inappropriate and sensitive content is important information that broadcasters need to know to inform viewers.
Also for brands and advertisers it is vital to associate their identity with appropriate content. We developed a Sensitive content detection module which is able to understand when sensitive content is shown.
Our module is trained to detect nudity, weapons and common sensitive content. We can train our module to detect specific sensitive content that needs to be found and flagged.
Metaliquid core combines visual and audio information to leverage the depth of semantic analysis and integrate the information in an overall output.
Our audio classification module can detect when there's a dialogue and the related language, when music is played and identify the related genre and when there is noise or specific sounds from the environment.
Determining where the action is taking place is precious information which requires a lot of time to be classified manually.
Our Setting detection module has been developed to answer this need and is able to identify in real time what the setting is and where each scene is shot.
Metaliquid can detect whether an action is taking place in an urban exterior, in a forest, in a car or in a room.
One of the most frequent viewers behavioral trends in the VOD market is binge watching.
Users, especially when watching series want to start a new episode as soon as they finish one. We developed a specific module to support content discovery features that suggest content to users, our binge watching module is trained to detect the title sequence and the ending credits scenes, enabling high efficiency in the management of binge watching functionality.