Project description
Digital human interaction for immersive communication
In the evolving digital world, how people communicate and perceive each other is changing rapidly. Emerging technologies allow humans to be digitised, creating digital doubles that exist in virtual and mixed reality spaces. These 3D digital humans are especially important for immersive telepresence, allowing real-time interactions in augmented or virtual environments. However, achieving realistic and interactive digital humans with consumer-grade hardware remains a challenge. The ERC-funded LeMo project tackles this by exploring the use of natural language to enhance the creation, modelling, and interaction of digital humans. Focusing on the upper body, the project aims to capture and synthesise lifelike motions, especially facial gestures and hand movements, for seamless and realistic communication in virtual environments.
Objective
We are taking part in a revolution of how people perceive each other and how they communicate in the digital space. Emerging technologies demonstrate how humans can be digitized such that their digital doubles (digital humans) can be viewed in mixed or virtual reality applications. Especially, for immersive telepresence these 3D digital humans are of high interest since people can see each other in augmented (AR) or virtual reality (VR), and can interact with each other. For an immersive experience this requires a high-quality level of appearance reproduction as well as the estimation of subtle motions of the human. Besides quality aspects, digital humans need to be reconstructed and driven by consumer-grade hardware solutions (e.g. a webcam) to enable a wide applicability of digital humans for immersive telecommunication, virtual mirrors (e-commerce), or other entertainment purposes (e.g. computer games). In this proposal, we concentrate on the computer vision and graphics aspects of capturing and rendering (realizing) motions and appearances of photo-realistic digital humans reconstructed with commodity hardware and ask the question:
Can we leverage natural language for the reconstruction, representation, and modelling of the appearance, motion, and interactions of digital humans?
As the face and gestures play an important role during conversations, we focus our research on the upper body of the human, aiming at capturing and synthesizing complete, contact-rich digital humans with life-like motions. We will analyze how humans move and will develop data-driven motion synthesis methods that learn how to generate talking and listening motion with interactions of the hand with the face.
Fields of science (EuroSciVoc)
CORDIS classifies projects with EuroSciVoc, a multilingual taxonomy of fields of science, through a semi-automatic process based on NLP techniques. See: https://op.europa.eu/en/web/eu-vocabularies/euroscivoc.
CORDIS classifies projects with EuroSciVoc, a multilingual taxonomy of fields of science, through a semi-automatic process based on NLP techniques. See: https://op.europa.eu/en/web/eu-vocabularies/euroscivoc.
- social sciencespolitical sciencespolitical transitionsrevolutions
- social scienceseconomics and businessbusiness and managementcommercee-commerce
You need to log in or register to use this function
Keywords
Programme(s)
- HORIZON.1.1 - European Research Council (ERC) Main Programme
Topic(s)
Funding Scheme
HORIZON-ERC - HORIZON ERC GrantsHost institution
64289 Darmstadt
Germany