AI-driven Facial, Character & Speech Recognition


Trophées SATIS - Prix Futur et R&D



Broadcasters and content producers are now having to manage Petabytes of content data every month. As well as storing existing and recent content on-premise, broadcasters have also been digitising and archiving older analogue content, much of which still has a financial value as well as having wider cultural signifiance. It is therefore extremely important that content is appropriately catalogued using as many different data points as possible, and that these catalogues are easily searchable, enabling broadcasters and producers to quickly identify and retrieve their valuable video assets. In order to facilitate this process, Limecraft is introducing powerful AI-driven facial, character (text) and speech recognition options that help identify content by specific actor and/or by characters and text that appear in a scene or on the screen somewhere and/or by specific phrases, fragments of speech and dialogue.

Performances et avantages
The benefits of this new functionality relate to speed of retrieval of content, accuracy of search results and efficiency of time spent searching for content. As an example, this new functionality would enable an archivist to generate a short-list of all scenes within a catalogue/archive/asset management system featuing a specific actor, e.g. Jean-Paul Belmondo, and specific text appearing on-screen, e.g. a brand of beer appearing on a bottle in his hand or on a table. This functionality is very clearly going to speed up the search/retreival process, as well as providing additional data-points that will make the search process far more powerful. The AI engine is capable of identifying actors throughout the different stages of their career, regardless of age, and provides extremely accurate results regardless of gender or ethnicity.

Date première
Will be beta-launched at SATIS