Making the journey of learning and discovery as exciting as possible.
...make education more engaging and improve learning outcomes by giving children the freedom to discover educational content themselves?
...leverage a world-class back catalogue of 400,000 hours of archival footage that is currently sitting unused?
...combining machine learning and virtual reality to create a learning platform where children can watch educational content and follow their innate curiosity to explore ideas.
Unity; Concept extraction; Virtual reality; Three.js
A collaboration with the BBC Archives team resulted in creating a proof-of-concept immersive learning platform for exploring and visualising over 400,000 hours of archive footage. Unstructured subtitle text was analysed to create contextual links between specific clips within videos. These clips were then rendered in a procedurally generated landscape that could be explored in real time. By leveraging machine learning to tag video footage with rich metadata on a minute-by-minute basis, we could unlock new ways of finding and linking content in an otherwise unwieldy back catalogue.
ML extracted concepts highlight key points for each video and link to related content
The platform was targeted at young learners who can easily become disengaged with traditional learning approaches. Through the self-guided exploration of educational content, “active learning” is encouraged by giving children the freedom to discover new ideas and follow the paths that they find most interesting.
Tangible features like “footprints” allow users to track progress and see where others have travelled before
Individual video clips were placed within procedurally generated landscapes driven by the extracted concepts from each video. This rich visual backdrop, combined with spatial audio, provided the immersive experience that fuelled the desire for exploration and complimented the footage being watched.
Learn by travelling to the ends of the Earth
Additional information#