Accurate 3D Scene Reconstruction From Images With Neural Method H/F

CEA; Saclay; Sur place; EUR 40 000 - 60 000; Plein temps

Résumé du poste

Une organisation de recherche innovante basée à Saclay propose un stage sur la reconstruction 3D de scènes à partir d'images. Les candidats devront concevoir une méthode de reconstruction en utilisant des techniques d'apprentissage profond telles que Neural Fields. Ce stage offre des opportunités de recherche autonome et la possibilité de poursuivre un doctorat après le stage.

Prestations

Stipend between €1300 and €1400 per month

75% reimbursement on public transportation costs

Opportunity to continue with a PhD or as a research engineer

Qualifications

• Étudiant(e) en 5ème année ou en année de césure.
• Maîtrise de Python dans un cadre d'apprentissage profond (de préférence PyTorch).

Responsabilités

• Explorer les techniques de pointe pour la reconstruction 3D à partir d'images 2D.
• Concevoir et développer une méthode de reconstruction de surfaces 3D.
• Implémenter une approche existante dans le cadre de Neural Fields.

Connaissances

Python proficiency

Deep learning frameworks (preferably PyTorch)

Formation

Students in their 5th year of studies (Master 2) or gap year

Description du poste

Accurate 3D scene reconstruction from images with neural Method H/F

Référence

2025-37890

Based in Saclay (Essonne), the LIST is one of the two institutes of CEA Tech, the Technological Research Division of the CEA. Dedicated to intelligent digital systems, its mission is to carry out technological developments of excellence on behalf of industrial partners, in order to create value.

Within the LIST, the Laboratory of Vision for Modeling and Localization (LVML) conducts its research in the field of computer vision and artificial intelligence for the perception of intelligent and autonomous systems. The laboratory's research themes include 3D localization, segmentation, characterization and vision for robotics.

Domaine

Contrat

Stage

Accurate 3D scene reconstruction from images with neural Method H/F

Stage 3D reconstruction of an environment or object using multiple 2D images observed at different viewpoints is a well‑studied research topic with multiple applications in robotics, medicine, AR/VR, and automotive. Recent work using AI techniques so‑called Neural Fields have revolutionized the field of rendering‑based images. While generating highly realistic visualization of 3D environments, Neural Fields also recover dense 3D surfaces leading to high‑fidelity 3D reconstruction. However, those techniques have limitations in terms of robustness and accuracy when it comes to large environments and specular objects such as metallic surfaces in an industrial environment.

Missions during this internship, you will explore the state‑of‑the‑art techniques for 3D scene reconstruction from 2D image using Neural Fields. With a focus on the accuracy of reconstructed geometry, you will design and develop a method of 3D surface reconstruction, which will be applied to industrial environment.

The internship will notably include:

• Reviewing existing literature on the topic
• Selecting and implementing an existing approach in the Neural Fields framework NeRFStudio
• Benchmarking the new method using images capturing an industrial environment

Job‑related benefits

Joining the CEA List and the LVML as an intern means:

• Working in one of the most innovative research organizations in the world, addressing societal challenges to build the world of tomorrow
• Discovering a rich ecosystem: privileged connections between the industrial and academic sectors
• Conducting research autonomously and creatively: encouragement to publish results (scientific articles, patents, open‑source codes…)
• Join a young and dynamic team
• Benefit from an internal computing infrastructure with more than 300 state‑of‑the‑art GPUs
• Receive a stipend between €1300 and €1400 per month
• Have the opportunity to continue with a PhD or as a research engineer after the internship
• Receive a 75% reimbursement on public transportation costs, and benefit from the “mobili‑jeune” aid to reduce rent costs…

Qualifications

• Students in their 5th year of studies (Master 2) or gap year
• Python proficiency in a deep learning framework (preferably PyTorch)