PhD CIFRE, PhD – Next-generation realistic avatar representation for videoconferencing
Core
Develop Gaussian-avatar representations for multi-user immersive telepresence, extending avatar coding from face-centric to unified full-body and facial representations.
Role type
Research PhD (CIFRE) in computer vision and real-time avatar communication
Builds
Compact, semantic, and standard-compatible avatar representations for videoconferencing and metaverse applications
Domain
AI, computer vision, wireless communications, video processing
Deliverable
production ML models
Required skills
3D Gaussian Splatting, geometric priors, view-dependent rendering, skeletal animation, blendshape control, real-time streaming optimization
Preferred skills
multi-user virtual reality, semantic data solutions, low-latency transmission protocols
Technologies
3DGS, GaussianBlendshape, GaussianAvatars, FlashAvatar, HUGS
Responsibilities
Investigate geometry-aware Gaussian representations for realistic 3D head models; Develop semantic control mechanisms for animating Gaussians via skeletal joints and blendshape weights; Evaluate quality of experience for full-body and facial encoding/decoding in shared applications
Seniority
PhD candidate, research-focused
