Post-Doctorant Hybrid Representations And Adaptive Streaming Of 4D Gaussians For Dynamic Scenes Hybrid Representations And Adaptive Streaming Of 4D Gaussians For Dynamic Scenes H/F

INRIA

Post-Doctorant F/H Hybrid Representations and Adaptive Streaming of 4D Gaussians for Dynamic Scenes

Type de contrat : CDD

Niveau de diplôme exigé : Thèse ou équivalent

Fonction : Post-Doctorant

A propos du centre ou de la direction fonctionnelle

The Inria Centre at Rennes University is one of Inria's eight centres and has more than thirty research teams. The Inria Centre is a major and recognized player in the field of digital sciences. It is at the heart of a rich R&D and innovation ecosystem: highly innovative PMEs, large industrial groups, competitiveness clusters, research and higher education players, laboratories of excellence, technological research institute, etc.

Principales activités

Hybrid Representations and Adaptive Streaming of 4D Gaussians for Dynamic Scenes

Host Institution:Inria Centre at the University of Rennes (teams Combo and Compact)
Industrial Partner:InterDigital R&D France (Cesson-Sévigné / Rennes)
Duration:18 months
Inria Supervisors:Christine Guillemot / Pierre Hellier (email:)
InterDigital Co-supervisors:Neus Sabater / Gustavo Sandri (email:)

Context and Collaborative Framework

The Inria research team and InterDigital France (formerly Technicolor) share a long-standing and fruitful collaboration in the fields of computer vision and immersive data compression.

This postdoctoral position is a direct connection with a recent jointly-supervised Ph.D. thesis focusing on the "Generation and Compression of Dynamic 3D Gaussians". While the Ph.D. focuses on the groundwork for compressing 3D Gaussian Splatting (3DGS) models, the postdoc shall explore initial solutions for temporal sequences (4D).

The postdoctoral researcher will be officially employed by Inria, but the project will be conducted in a hybrid immersion, featuring continuous interaction and frequent stays at InterDigital's labs. This format aims to combine the excellence of Inria's academic research with InterDigital's industrial strength and standardization expertise (MPEG).

Motivations and State-of-the-Art

3D Gaussian Splatting(3DGS) has emerged as a major alternative to NeRFs for Novel View Synthesis (NVS), offering unprecedented visual quality and real-time rendering. However, representing dynamic scenes (4DGS) leads to an explosion in memory footprint, drastically limiting its use in real-world applications (AR/VR, telepresence, immersive video streaming, etc.).

The Ph.D. thesis, that started in January 2025, is developing promising compression schemes based on pruning, vector quantization, and exploiting structural redundancies.

To scale up to consumer applications, it is now necessary to go beyond simple "offline" compression. We need to design 4D Gaussian models natively built for adaptive transmission (streaming) and explore hybrid representations that couple splats with the knowledge of 3D objects in motion, to optimize the rate-distortion tradeoff.

Objectives and Research Directions

The postdoctoral researcher will be tasked with pushing the boundaries of immersive 4D representations. The envisioned research directions are:

• Axis 1: Hybrid Representations (Surfaces + 3D/4D Gaussians)
  The most costly part of 3DGS to be encoded is the geometry. Hence, coupling 3D/4DGS with surface representations (meshes) offers major potential. The goal will be to use low-resolution meshes (which are very cheap to compress using existing MPEG standards) as a base structure upon which residual dynamic Gaussians will be anchored.
• Axis 2: Adaptive Streaming and Temporal Level of Detail (LoD)
  Design a hierarchical compression scheme for 4DGS allowing for progressive decoding and view-dependent rendering/streaming. This will involve optimizing real-time bandwidth allocation for massive dynamic sequences.
• Axis 3: Normalization and Standardization (MPEG)
  In direct collaboration with InterDigital teams, translate the algorithmic advances of the postdoc into Proofs of Concept (PoC) compatible with ongoing efforts within MPEG standardization groups (especially regarding upcoming 3D/Volumetric formats).

Expected Valorization

The conducted research will have a dual valorization goal, reflecting the Inria/InterDigital synergy:

• Academic:Publications in top-tier computer vision and AI conferences and journals (e.g., CVPR, ICCV, ECCV, SIGGRAPH, IEEE TPAMI).
• Industrial:Joint patent applications and direct technical contributions to international video/3D compression standards (MPEG).

Candidate Profile

• Degree:Recently defended Ph.D. in Computer Science, Computer Vision, or Signal Processing.
• Technical Skills:
• Proven expertise in Neural Rendering, Novel View Synthesis (NeRFs, 3D Gaussian Splatting).
• Solid knowledge in data compression (point clouds, video, generative models) is a major plus.
• Excellent proficiency in Python, PyTorch, proficiency in C++ is a plus.
• Soft Skills:Scientific rigor, ability to publish in top-tier conferences, strong interest in applied research for industrial standards, and a high capacity to work within a hybrid academic/industrial team.

Application Process

Interested candidates should send the following documents:

• A detailed CV including a full list of publications.
• A short cover letter explaining the candidate's fit with the research topic.
• The Ph.D. manuscript (or a link to it) and the thesis defense reports (if available).
• Contact information for 2 references.

Avantages

• Restauration subventionnée
• Transports publics remboursés partiellement
• Congés: 7 semaines de congés annuels + 10 jours de RTT (base temps plein) + possibilité d'autorisations d'absence exceptionnelle (ex : enfants malades, déménagement)
• Possibilité de télétravail (après 6 mois d'ancienneté) et aménagement du temps de travail
• Équipements professionnels à disposition (visioconférence, prêts de matériels informatiques, etc.)
• Prestations sociales, culturelles et sportives (Association de gestion des oeuvres sociales d'Inria)
• Accès à la formation professionnelle
• Sécurité sociale

Rémunération

Monthly gross salary from 2 788 euros.