2025Activity reportProject-TeamMORPHEO
RNSR: 201120981M- Research center Inria Centre at Université Grenoble Alpes
- In partnership with:CNRS, Université de Grenoble Alpes
- Team name: Capture and Analysis of Shapes in Motion
- In collaboration with:Laboratoire Jean Kuntzmann (LJK)
Creation of the Project-Team: 2014 January 01
Each year, Inria research teams publish an Activity Report presenting their work and results over the reporting period. These reports follow a common structure, with some optional sections depending on the specific team. They typically begin by outlining the overall objectives and research programme, including the main research themes, goals, and methodological approaches. They also describe the application domains targeted by the team, highlighting the scientific or societal contexts in which their work is situated.
The reports then present the highlights of the year, covering major scientific achievements, software developments, or teaching contributions. When relevant, they include sections on software, platforms, and open data, detailing the tools developed and how they are shared. A substantial part is dedicated to new results, where scientific contributions are described in detail, often with subsections specifying participants and associated keywords.
Finally, the Activity Report addresses funding, contracts, partnerships, and collaborations at various levels, from industrial agreements to international cooperations. It also covers dissemination and teaching activities, such as participation in scientific events, outreach, and supervision. The document concludes with a presentation of scientific production, including major publications and those produced during the year.
Keywords
Computer Science and Digital Science
- A5.1.8. 3D User Interfaces
- A5.3.4. Registration
- A5.5.1. Geometrical modeling
- A5.5.4. Animation
- A6.2.8. Computational geometry and meshes
- A9.2.6. Neural networks
- A9.2.8. Deep learning
- A9.12.4. 3D and spatio-temporal reconstruction
- A9.12.5. Object tracking and motion analysis
- A9.12.8. Motion capture
Other Research Topics and Application Domains
- B1.1.11. Plant Biology
- B2.7.2. Health monitoring systems
- B2.8. Sports, performance, motor skills
- B9.2.2. Cinema, Television
- B9.2.3. Video games
- B9.4. Sports
- B9.5.1. Computer science
- B9.5.2. Mathematics
- B9.7. Knowledge dissemination
- B9.7.1. Open access
- B9.7.2. Open data
1 Team members, visitors, external collaborators
Research Scientists
- Jean Franco [Team leader, INRIA, Senior Researcher, HDR]
- Stefanie Wuhrer [INRIA, Researcher, HDR]
Faculty Member
- Sergi PUJADES [UGA, Associate Professor, HDR]
PhD Students
- Youssef Ben Cheikh [INRIA]
- Antoine Dumoulin [INRIA]
- Vicente Estopier Castillo [INSERM]
- Samara Ghrer [INRIA]
- Felix Küper [UGA, from Oct 2025]
- Aymen Merrouche [UGA, ATER]
- Laura Neschen [INRIA]
- Nampoina Ravelomanana [INRIA]
- Rim Rekik Dit Nekhili [INRIA, until Jan 2025]
- Briac Toussaint [INRIA, until Mar 2025]
- Quentin Zoppis [UGA]
Technical Staff
- Vaibhav Arora [INRIA, Engineer, until Jan 2025]
- Mohammed Chekroun [INRIA, Engineer, until Feb 2025]
- Abdelmouttaleb Dakri [INRIA, Engineer, until Sep 2025]
- Gautier Marcon [INRIA, Engineer, from Sep 2025]
- Julien Pansiot [INRIA, Engineer]
Interns and Apprentices
- Mathias Petey [INRIA, Intern, from Mar 2025 until Sep 2025]
- Bastien Roure [INRIA, Intern, from May 2025 until Jul 2025]
- Bastien Roure [INRIA, Intern, until Apr 2025]
Administrative Assistants
- Julia Di Toro [INRIA]
- Nathalie Gillot [INRIA]
External Collaborators
- Felix Küper [ENSIMAG, until Feb 2025]
- Tomas Svaton [ENSIMAG, from Apr 2025 until Sep 2025]
2 Overall objectives
Dynamic Geometry Modeling: a human is captured while running. The different temporal acquisition instances are shown on the same figure.
MORPHEO's ambition is to perceive and interpret shapes that move using multiple camera systems. Departing from standard motion capture systems, based on markers, that provide only sparse information on moving shapes, multiple camera systems allow dense information on both shapes and their motion to be recovered from visual cues. Such ability to perceive shapes in motion brings a rich domain for research investigations on how to model, understand and animate real dynamic shapes, and finds applications, for instance, in gait analysis, bio-metric and bio-mechanical analysis, animation, games and, more insistently in recent years, in the virtual and augmented reality domain. The MORPHEO team particularly focuses on four different axes within the overall theme of 3D dynamic scene vision or 4D vision:
- Shape and appearance models: how to build precise geometric and photometric models of shapes, including human bodies but not limited to, given temporal sequences.
- Dynamic shape vision: how to register and track moving shapes, build pose spaces and animate captured shapes.
- Inside shape vision: how to capture and model inside parts of moving shapes using combined color and X-ray imaging.
- Shape animation: Morpheo is actively investigating animation acquisition and parameterization methodologies for efficient representation and manipulability of acquired 4D data.
The strategy developed by Morpheo to address the mentioned challenges is based on methodological tools that include in particular learning-based approaches, geometry, Bayesian inference and numerical optimization. In recent years, and following many successes in the team and the computer vision community as a whole, a particular effort is ongoing in the team to investigate the use of machine learning and neural learning tools in 4D vision.
3 Research program
3.1 Shape and Appearance Modeling
Standard acquisition platforms, including commercial solutions proposed by companies such as Microsoft, 3dMD or 4DViews, now give access to precise 3D models with geometry, e.g. meshes, and appearance information, e.g. textures. Still, state-of-the-art solutions are limited in many respects: They generally consider limited contexts and close setups with typically at most a few meter side lengths. As a result, many dynamic scenes, even a body running sequence, are still challenging situations; They also seldom exploit time redundancy; Additionally, data driven strategies are yet to be fully investigated in the field. The MORPHEO team builds on the Kinovis platform for data acquisition and has addressed these issues with, in particular, contributions on time integration, in order to increase the resolution for both shapes and appearances, on representations, as well as on exploiting machine learning tools when modeling dynamic scenes. Our originality lies, for a large part, in the larger scale of the dynamic scenes we consider as well as in the time super resolution strategy we investigate. Another particularity of our research is a strong experimental foundation with the multiple camera Kinovis platforms.
3.2 Dynamic Shape Vision
Dynamic Shape Vision refers to research themes that consider the motion of dynamic shapes, with e.g. shapes in different poses, or the deformation between different shapes, with e.g. different human bodies. This includes for instance shape tracking and shape registration, which are themes covered by MORPHEO. While progress has been made over the last decade in this domain, challenges remain, in particular due to the required essential task of shape correspondence that is still difficult to perform robustly. Strategies in this domain can be roughly classified into two categories: (i) data driven approaches that learn shape spaces and estimate shapes and their variations through space parameterizations; (ii) model based approaches that use more or less constrained prior models on shape evolutions, e.g. locally rigid structures, to recover correspondences. The MORPHEO team is substantially involved in both categories. The second one leaves more flexibility for shapes that can be modeled, an important feature with the Kinovis platform, while the first one is interesting for modeling classes of shapes that are more likely to evolve in spaces with reasonable dimensions, such as faces and bodies under clothing. The originality of MORPHEO in this axis is to go beyond static shape poses and to consider also the dynamics of shape over several frames when modeling moving shapes, and in particular with shape tracking and animation.
3.3 Inside Shape Vision
Another research axis is concerned with the ability to perceive inside moving shapes. This is a more recent research theme in the MORPHEO team that has gained importance. It was originally the research associated to the Kinovis platform installed in the Grenoble Hospitals. This platform is equipped with two X-ray cameras and ten color cameras, enabling therefore simultaneous vision of inside and outside shapes. We believe this opens a new domain of investigation at the interface between computer vision and medical imaging. Interesting issues in this domain include the links between the outside surface of a shape and its inner parts, especially with the human body. These links are likely to help understanding and modeling human motions. Until now, numerous dynamic shape models, especially in the computer graphics domain, consist of a surface, typically a mesh, rigged to a skeletal structure that is never observed in practice but allows to parameterize human motion. Learning more accurate relationships using observations can therefore significantly impact the domain.
3.4 Shape Animation
3D animation is a crucial part of digital media production with numerous applications, in particular in the game and motion picture industry. Recent evolutions in computer animation consider real videos for both the creation and the animation of characters. The advantage of this strategy is twofold: it reduces the creation cost and increases realism by considering only real data. Furthermore, it allows to create new motions, for real characters, by recombining recorded elementary movements. In addition to enable new media contents to be produced, it also allows to automatically extend moving shape datasets with fully controllable new motions. This ability appears to be of great importance with deep learning techniques and the associated need for large learning datasets. In this research direction, we investigate how to create new dynamic scenes using recorded events. More recently, this also includes applying machine learning to datasets of recorded human motions to learn motion spaces that allow to synthesize novel realistic animations.
4 Application domains
4.1 4D modeling
Modeling shapes that evolve over time, analyzing and interpreting their motion has been a subject of increasing interest of many research communities including the computer vision, the computer graphics and the medical imaging communities. Recent evolutions in acquisition technologies including 3D depth cameras (Time-of-Flight and Kinect), multi-camera systems, marker based motion capture systems, ultrasound and CT scanners have made those communities consider capturing the real scene and their dynamics, create 4D spatio-temporal models, analyze and interpret them. A number of applications including dense motion capture, dynamic shape modeling and animation, temporally consistent 3D reconstruction, motion analysis and interpretation have therefore emerged.
4.2 Shape Analysis
Most existing shape analysis tools are local, in the sense that they give local insight about an object's geometry or purpose. The use of both geometry and motion cues makes it possible to recover more global information, in order to get extensive knowledge about a shape. For instance, motion can help to decompose a 3D model of a character into semantically significant parts, such as legs, arms, torso and head. Possible applications of such high-level shape understanding include accurate feature computation, comparison between models to detect defects or medical pathologies, and the design of new biometric models.
4.3 Human Motion Analysis
The recovery of dense motion information enables the combined analysis of shapes and their motions. This allows to classify based on motion cues, which can help in the identification of pathologies or the design of new prostheses. Analyzing human motion can also be used to learn generative models of realistic human motion, which is a recent research topic within Morpheo.
5 Social and environmental responsibility
5.1 Footprint of research activities
The footprint of our research activity is dominated by dissemination and data costs.
Dissemination strategy: Traditionally, Morpheo's dissemination strategy has been to publish in the top conferences of the field (CVPR, ECCV, ICCV, MICCAI), as well as to give invited talks, leading to some long disance work trips. We currently try to increase submissions to top journals in the field directly, hence reducing travels, while still attending some in-person conferences or seminars to allow for networking.
Data management: The data produced by the Morpheo team occupies large memory volumes. The Kinovis platform typically produces around 1.5GB per second when capturing one actor at 50fps. The platform that also captures X-ray images at CHU produces around 1.3GB of data per second at 30fps for video and X-rays. For practical reasons, we reduce the data as much as possible with respect to the targeted application by only keeping e.g. 3D reconstructions or down-sampled spatial or temporal camera images. Yet, acquiring, processing and storing these data is costly in terms of resources. In addition, the data captured by these platforms are personal data with highly constrained regulations. Our data management therefore needs to consider multiple aspects: data encryption to protect personal information, data backup to allow for reproducibility, and the environmental impact of data storage and processing. For all these aspects, we are constantly checking for new and more satisfactory solutions.
For data processing, we rely heavily on cluster uses, in particular with neural networks which are known to have a heavy carbon footprint. Yet, in our research field, these types of processing methods have been shown to lead to significant performance gains. For this reason, we continue to use neural networks as tools, while attempting to use architectures that allow for faster and more energy efficient training, and simple test cases that can often be trained on local machines (with GPU) to allow for less costly network design. A typical example of this type of research is the work of Briac Toussaint, whose first thesis contribution exhibits a high quality reconstruction algorithm for the Kinovis platform, with under a minute of GPU computation time per frame, an order of magnitude faster than our previous research solution, yet it achieves improved precision.
5.2 Impact of research results
Morpheo's main research topic is not related to sustainability. Yet, some of our research directions could be applied to solve problems relevant to sustainability.
Realistic digital human modeling holds the potential to allow for realistic interactions through large geographic distances, leading to more realistic and satisfactory teleconferencing systems. Morpheo captures and analyzes humans in 4D, thereby allowing to capture both their shapes and movement patterns, and actively works on human body modeling. In this line of work, Morpheo participates through the 3DMOVE project and the Nemo.AI joint lab with Interdigital on advancing this topic.
Of course, as with any research direction, ours can also be used to generate technologies that are resource-hungry and whose necessity may be questionable, such as inserting animated 3D avatars in scenarios where simple voice communication would suffice, for instance.
6 Highlights of the year
6.1 Accepted projects
Program Inria Quadrant: Kinoflex
Under the impulse of Julien Pansiot, a PIQ project has been accepted toward the year 2026 activity, to conceive cost-efficient, lightweight, custom multi-camera capture plateforms.
BPI: Fabric
A project has been accepted toward a 2026 kickoff, to build a volumetric capture ecosystem (capture platform, educational support, asset library), with Morpheo as provider of new riggable volumetric capture algorithms and solutions. The project includes several volumetric capture industry partners, such as Novaya who leads the project, Ecole Melies for digital arts education, and IRCAM as immersive audio capture partner.
6.2 Organized Events
Stefanie Wuhrer co-organized Dagstuhl seminar Generative Models for 3D Vision, which took place May 11 – May 16, 2025 in Dagstuhl, Germany.
7 Latest software developments, platforms, open data
7.1 Latest software developments
7.1.1 Millimetric humans
-
Name:
Millimetric Human Surface Capture in Minutes
-
Keywords:
Multi-View reconstruction, Differentiable Rendering, Recognition of human movement
-
Scientific Description:
Detailed human surface capture from multiple images is an essential component for many 3D production, analysis and transmission tasks. Yet producing millimetric precision 3D models in practical time, and actually verifying their 3D accuracy in a real-world capture context, remain key challenges due to the lack of specific methods and data for these goals. We propose two complementary contributions to this end. The first one is a highly scalable neural surface radiance field approach able to achieve millimetric precision by construction, while demonstrating high compute and memory efficiency. The second one is a novel dataset of clothed mannequin geometry captured with a high resolution hand-held 3D scanner paired with calibrated multi-view images, that allow to verify the millimetric accuracy claim.
Although our approach can produce such a highly dense and precise geometry, we show how aggressive sparsification and optimizations of the neural surface pipeline lead to estimations requiring only minutes of computation time and a few GB of VRAM memory on GPU, while allowing for real-time millisecond neural rendering. On the basis of our framework and dataset, we provide a thorough experimental analysis of how such accuracies and efficiencies are achieved in the context of multi-camera human acquisition.
-
Functional Description:
The code allows to reconstruct geometry from human multi-view capture datasets.
- URL:
- Publication:
-
Contact:
Jean Franco
-
Participants:
Briac Toussaint, Jean Franco
7.1.2 HIT
-
Name:
Human Implicit Tissues
-
Keywords:
Implicit surface, Geometry
-
Functional Description:
This software takes as input the shape of the surface of a person, in the SMPL parametrization, and predicts the volumetric tissues (lean tissue, adipose tissue and bone tissue) inside it.
- URL:
-
Contact:
Sergi Pujades
-
Partner:
Max Planck Institute for Intelligent Systems
7.1.3 SKELJ
-
Name:
On predicting 3D bone locations inside the human body
-
Keywords:
Human Body Surface, Shape approximation, 3D, 3D modeling
-
Functional Description:
This software takes as input the shape of the surface of a person, in the SMPL parametrization, and outputs the 3D location of the bones, parametrized as a the SKEL mesh.
- URL:
-
Contact:
Sergi Pujades
-
Partner:
Max Planck Institute for Intelligent Systems
7.1.4 Osatta
-
Name:
One-Shot Automatic Test Time Augmentation for Domain Adaptation
-
Keywords:
Image analysis, Image processing, Machine learning
-
Functional Description:
This software computes a family of image transformations that allow to improve the accuracy of a fundamental model on biased cohorts with few annotations.
- URL:
-
Contact:
Sergi Pujades
7.1.5 MotionRetargeter
-
Name:
Correspondence-free online human motion retargeting
-
Keywords:
3D animation, 3D modeling
-
Functional Description:
Allows to retarget the motion of one human body to another body shape, without needing any correspondence information. Code allows to reproduce results in: Mathieu Marsot, Rim Rekik, Stefanie Wuhrer, Jean-Sébastien Franco, Anne-Hélène Olivier. Correspondence-free online human motion retargeting. arXiv 2302.00556, 2023.
- URL:
-
Contact:
Rim Rekik Dit Nekhili
7.1.6 ProbeSDF
-
Name:
ProbeSDF: Light Field Probes for Neural Surface Reconstruction
-
Keywords:
3D reconstruction, Differentiable Rendering, Multi-View reconstruction
-
Functional Description:
The code allows to reconstruct geometry from human multi-view capture datasets.
- URL:
-
Contact:
Jean Franco
-
Participants:
Jean Franco, Briac Toussaint
7.1.7 SKEL
-
Name:
From Skin to Skeleton: Towards Biomechanically Accurate 3D Digital Humans
-
Keywords:
Anatomy, 3D modeling, 3D movement
-
Functional Description:
The software allows to generate 3D biomechanical skeletons from 3D skin surface observations. This software is associated with the publication "From Skin to Skeleton: Towards Biomechanically Accurate 3D Digital Humans" Sigg Asia 2023 (https://skel.is.tue.mpg.de)
- URL:
-
Contact:
Sergi Pujades
7.1.8 KinovisReconstruction
-
Name:
Differential rendering based surface reconstruction for the Kinovis platform
-
Keywords:
3D, 4D, Computer vision, Differentiable Rendering, 3D reconstruction, Multi-View reconstruction
-
Functional Description:
Very fast reconstruction and rendering software based on neural inverse differential rendering.
- URL:
-
Contact:
Briac Toussaint
7.1.9 KinovisCalibration
-
Name:
Multi-view calibration software for the Kinovis platform
-
Keywords:
3D, Computer vision, Calibration
-
Scientific Description:
The software is the result of the engineering master of Briac Toussaint and offers a calibration suite bringing together a set of classical techniques (point detection, epipolar geometry estimation, epipolar geometry models extended to radial distortion, RANSAC, optimization graph on all views, auto-calibration) for the case of complete calibration of a multi-view image set, tested in particular on the Kinovis platform.
-
Functional Description:
Multi-camera calibration software specialized in the management of large multi-camera platforms, based on a T-shaped calibration stick.
- URL:
-
Contact:
Julien Pansiot
-
Participants:
Briac Toussaint, Julien Pansiot, Jean Franco
7.1.10 KinovisEnvironment
-
Name:
Video and 3D data processing environment
-
Keywords:
3D, Videos
-
Functional Description:
Complete environment to automatically process data from the Kinovis platform: copy, demosaicing, camera calibration, reconstruction, post-processing, verification, archiving, etc.
-
Contact:
Julien Pansiot
7.1.11 Medical Image Registration Pipeline
-
Name:
Medical Image Registration piepline
-
Keywords:
3D, Medical imaging, Image segmentation, Image registration
-
Functional Description:
The repository provides a pipeline for medical image preprocessing, rigid and non-rigid alignment. It includes steps for background removal, bone and envelope generation, and image alignment.
-
Contact:
Mohammed Chekroun
7.2 New platforms
Kinovis platform deployment.
Participants: Julien Pansiot, Aurelien Courvoisier, Sergi Pujades.
The Kinovis platform has been partly moved to the Grenoble Hospital (CHU), where it was used on patients by clinicians. This Proof-of-Concept aims to demonstrate the following points:
- the platform mobility at a technical level
- the ability to capture in varied environments at a scientific level
- the tremendous potential in real-life applications
The proof-of-concept platform offers promising prospects. During the year 2025, the data of 250 scoliosis patients has been collected and will be used for future research in the team and in collaboration with Aurélien Courvoisier (PUPH). It has served as testbed for the most recent algorithms in the team, such as ProbeSDF 7, in an entirely different capture scenario and camera configuration than the Kinovis studio in Montbonnot, to resconstruct the full geometry of patients, thereby providing a means to monitor patient shape and motion in the full 3D volume. The yearlong experiment has provided a strong basis for the installation of a permanent platform in the future, for which we are currently projecting to secure finding.
Kinoflex PIQ project.
Participants: Julien Pansiot, Gautier Marçon.
While Kinovis is a state of the art platform for volumetric capture and experimental purposes, it is mainly intended for prospective research and its hardware configuration is not easy to replicate or displace in a different setting. For this reason, The Kinoflex PIQ project is leading to applied research on the building of a new low cost and flexible, custom-built platform, with a sensor and harwdare configuration built from the ground up. The new proof of concept plateform intends to demonstrate the following :
- new platform mobility possibilities
- simplified and cost efficient deployment in a variety of environments
- new transfer opportunities based on compact and low cost design
The PIQ project has been accepted in 2025 and development has started end of the year, with results expected in the coming years.
8 New results
8.1 Efficient Finger Model and Accurate Tracking for Hover-and-Touch, Mid-air and Microgesture Interaction
Participants: Quentin Zoppis, Sergi Pujades, Laurence Nigay, François Bérard.
Use-case of finger tracking for interaction with a 3D scene.
Bare-handed gestural interaction with computer systems is widespread, whether with touchscreens or Augmented Reality headsets. Various forms of gestural interaction exist including hover-and-touch, mid-air and microgesture interaction. Studying the full benefits of these gestural interactions, and their combinations, is currently not possible due to the inadequate performances of the existing tracking solutions. To address this problem, we propose a marker-based visual tracking algorithm with a novel finger model, and its open source implementation. A key contribution is the simplicity of the finger and fingertip model (i.e. cylinders and a sphere respectively). This simple model leads to low computational cost (600 microseconds), high precision (0.02 mm) and accurate (one millimeter) fingertip tracking, without impeding finger movement. We illustrate the benefits of the proposed tracking approach with a demonstration application combining hover-and-touch, mid-air and microgesture interactions for editing a 3D point cloud.
This work was published at the UIST 2025 user interface conference, 8.
8.2 ProbeSDF: Light Field Probes For Neural Surface Reconstruction
Participants: Briac Toussaint, Diego Thomas, Jean-Sébastien Franco.
ProbeSDF pipeline and main results.
SDF-based differential rendering frameworks have achieved state-of-the-art multiview 3D shape reconstruction. In this work, we re-examine this family of approaches by minimally reformulating its core appearance model in a way that simultaneously yields faster computation and increased performance. To this goal, we exhibit a physically inspired minimal radiance parametrization decoupling angular and spatial contributions, by encoding them with a small number of features stored in two respective volumetric grids of different resolutions. Requiring as little as four parameters per voxel, and a tiny MLP call inside a single fully fused kernel, our approach allows to enhance performance with both surface and image (PSNR) metrics, while providing a significant training speedup and real-time rendering. We show this performance to be consistently achieved on real data over two widely different and popular application fields, generic object and human subject shape reconstruction, using four representative and challenging datasets
This work was published at the CVPR 2025 computer vision conference 7. Project page.
8.3 Quality assessment of 3D human animation: Subjective and objective evaluation
Participants: Rim Rekik, Stefanie Wuhrer, Ludovic Hoyet, Katja Zibrek, Anne-Hélène Olivier.
Quality assessment of 3D human animation
Virtual human animations have a wide range of applications in virtual and augmented reality. While automatic generation methods of animated virtual humans have been developed, assessing their quality remains challenging. Recently, approaches introducing task-oriented evaluation metrics have been proposed, leveraging neural network training. However, quality assessment measures for animated virtual humans that are not generated with parametric body models have yet to be developed. In this context, we introduce a first such quality assessment measure leveraging a novel data-driven framework. First, we generate a dataset of virtual human animations together with their corresponding subjective realism evaluation scores collected with a user study. Second, we use the resulting dataset to learn predicting perceptual evaluation scores. Results indicate that training a linear regressor on our dataset results in a correlation of 90%, which outperforms a state of the art deep learning baseline.
This work was published in IEEE Transactions on Visualization and Computer Graphics 5. DataProject page
8.4 VortSDF: 3D Modeling with Centroidal Voronoi Tessellation on Signed Distance Field
Participants: Diego Thomas, Briac Toussaint, Jean-Sébastien Franco, Edmond Boyer.
VortSDF teaser and main result.
Volumetric shape representations have become ubiquitous in multi-view reconstruction tasks. They often build on regular voxel grids as discrete representations of 3D shape functions, such as SDF or radiance fields, either as the full shape model or as sampled instantiations of continuous representations, as with neural networks. Despite their proven efficiency, voxel representations come with the precision versus complexity trade-off. This inherent limitation can significantly impact performance when moving away from simple and uncluttered scenes. In this paper we investigate an alternative discretization strategy with the Centroidal Voronoi Tessellation (CVT). CVTs allow to better partition the observation space with respect to shape occupancy and to focus the discretization around shape surfaces. To leverage this discretization strategy for multi-view reconstruction, we introduce a volumetric optimization framework that combines explicit SDF fields with a shallow color network, in order to estimate 3D shape properties over tetrahedral grids. Experimental results with Chamfer statistics validate this approach with unprecedented reconstruction quality on various scenarios such as objects, open scenes or human.
This work was published at the WACV 2025 computer vision conference 6. Project page.
8.5 MedShapeNet – a large-scale dataset of 3D medical shapes for computer vision
Participants: Sergi Pujades.
3D shapes from MedShapeNet
Objectives: The shape is commonly used to describe the objects. State-of-the-art algorithms in medical imaging are predominantly diverging from computer vision, where voxel grids, meshes, point clouds, and implicit surface models are used. This is seen from the growing popularity of ShapeNet (51,300 models) and Princeton ModelNet (127,915 models). However, a large collection of anatomical shapes (e.g., bones, organs, vessels) and 3D models of surgical instruments is missing.
Methods: We present MedShapeNet to translate data-driven vision algorithms to medical applications and to adapt state-of-the-art vision algorithms to medical problems. As a unique feature, we directly model the majority of shapes on the imaging data of real patients. We present use cases in classifying brain tumors, skull reconstructions, multi-class anatomy completion, education, and 3D printing.
Results: By now, MedShapeNet includes 23 datasets with more than 100,000 shapes that are paired with annotations (ground truth). Our data is freely accessible via a web interface and a Python application programming interface and can be used for discriminative, reconstructive, and variational benchmarks as well as various applications in virtual, augmented, or mixed reality, and 3D printing.
Conclusions: MedShapeNet contains medical shapes from anatomy and surgical instruments and will continue to collect data for benchmarks and applications.
This work was published at Biomedical Engineering/Biomedizinische Technik 70, no. 1 (2025): 71-90. 4. Project page
9 Bilateral contracts and grants with industry
9.1 Bilateral contracts with industry
Participants: Antoine Dumoulin, Youssef Ben Cheikh, Stefanie Wuhrer, Jean-Sébastien Franco.
- The Morpheo INRIA team has a collaboration with Interdigital in Rennes through the Nemo.AI joint lab. The kickoff to this collaboration was in November 2022. The collaboration involves two PhD co-supervisions, Antoine Dumoulin and Youssef Ben Cheikh, at Centre INRIA de l'Université Grenoble Alpes. The subject of the collaboration revolves around digital humans, one the one hand to estimate the clothing of humans from images, and on the other to estimate hair models from one or several videos.
10 Partnerships and cooperations
10.1 International initiatives
Morpheo is part of a submission application to create a Grenoble Ellis Unit, federating the Inria teams, researchers and UGA and private partners in the Grenoble area. As part of this process, all permanent members of the Morpheo team have applied to become Ellis members and been accepted.
10.2 National initiatives
10.2.1 ANR
ANR JCJC 3DMOVE - Learning to synthesize 3D dynamic human motion.
Participants: Stefanie Wuhrer.
The 3DMOVE project focused on developing tools to process and analyze raw dynamic human capture data in 4D robustly and automatically. 3DMOVE developed generative models of human motion, which can be leveraged to create plausible synthetic human motion sequences, that have the potential to influence virtual reality applications such as virtual change rooms or crowd simulations. Developing such tools was challenging due to the high variability in human shape and motion and due to significant geometric and topological acquisition noise present in state-of-the-art acquisitions. 3DMOVE leveraged recently developed deep learning techniques. The project also developed tools to assess the quality of the generated motions using perceptual studies. This project involved two Ph.D. students: Mathieu Marsot who graduated in May 2023 and Rim Rekik who graduated in June 2025. 3DMOVE ended in January 2025.
ANR Equipex+ CONTINUUM - Collaborative continuum from digital to human
Participants: Julien Pansiot, Jean-Sebastien Franco.
The CONTINUUM project will create a collaborative research infrastructure of 30 platforms located throughout France, including Inria Grenoble's Kinovis, to advance interdisciplinary research based on interaction between computer science and the human and social sciences. Thanks to CONTINUUM, 37 research teams will develop cutting-edge research programs focusing on visualization, immersion, interaction and collaboration, as well as on human perception, cognition and behaviour in virtual/augmented reality, with potential impact on societal issues. CONTINUUM enables a paradigm shift in the way we perceive, interact, and collaborate with complex digital data and digital worlds by putting humans at the center of the data processing workflows. The project will empower scientists, engineers and industry users with a highly interconnected network of high-performance visualization and immersive platforms to observe, manipulate, understand and share digital data, real-time multi-scale simulations, and virtual or augmented experiences. All platforms will feature facilities for remote collaboration with other platforms, as well as mobile equipment that can be lent to users to facilitate onboarding.
ANR PRC Inora
Participants: Sergi Pujades, Julien Pansiot.
The INORA project aims at understanding the mechanisms of action of shoes and orthotic insoles on Rheumatoid arthritis (RA) patients through patient-specific computational biomechanical models. These models will help in uncovering the mechanical determinants to pain relief, which will enable the long-term well-being of patients. Motivated by the numerous studies highlighting erosion and joint space narrowing in RA patients, we postulate that a significant contributor to pain is the internal joint loading when the foot is inflamed. This hypothesis dictates the need of a high-fidelity volumetric segmentation for the construction of the patient-specific geometry. It also guides the variables of interests in the exploitation of a finite element (FE) model. The INORA project aims at providing numerical tools to the scientific, medical and industrial communities to better describe the mechanical loading on diseased distal foot joints of RA patients and propose a patient-specific methodology to design pain-relief insoles. The Mines de St Etienne is leading this project.
ANR 4DPlants - Learning Plant Growth Models
Participants: Stefanie Wuhrer.
The overall goal of the 4DPlants project is to invent solutions that will make plant phenotyping in 3D and over time reliable and accessible to a majority of plant scientists. This is achieved in a consortium including biologists and computer scientists by acquiring growing plants in 3D, and by analyzing growth changes using data-driven methods. The project, which started in December 2024, includes partners from Université de Strasbourg, ENS Lyon, and Morpheo team at Inria. This projet involves Ph.D. student Samara Ghrer, hired in December 2024.
10.3 Regional initiatives
Persyval PhD Grant.
Participants: Quentin Zoppis, Sergi Pujades, Laurence Nigay, François Berard.
In collaboration with the IIHM team from the LIG, a co-supervision started on the topic of high precision real-time tracking of fingers with markerless vision systems.
11 Dissemination
Participants: Stefanie Wuhrer, Sergi Pujades, Jean-Sébastien Franco.
11.1 Promoting scientific activities
11.1.1 Scientific events: organisation
General chair, scientific chair
- Stefanie Wuhrer co-organized Dagstuhl seminar Generative Models for 3D Vision, which took place May 11 – May 16, 2025 in Dagstuhl, Germany
Reviewer
- Sergi Pujades reviewed for MICCAI and ODIN Workshop
- Jean-Sébastien Franco reviewed for CVPR, 3DV, Siggraph Asia
- Stefanie Wuhrer reviewed for CVPR, ICCV, 3DV, Siggraph, 3DOR
11.1.2 Journal
Member of the editorial boards
- Jean-Sébastien Franco was associate editor for the International Journal on Computer Vision.
Reviewer - reviewing activities
- Sergi Pujades reviewed for IJCV
- Stefanie Wuhrer reviewed for Computers & Graphics
11.1.3 Invited talks
- Sergi Pujades was invited as a keynote speaker at the 1st edition of the Workshop "From Scene Understanding to Human Modeling" held with BMVC 2025.
- Sergi Pujades was invited as a keynote speaker at LJK laboratory CNRS evaluation.
- Jean-Sébastien Franco gave a talk to the HCERES commitee evaluating the LJK laboratory, on 03.12.2025, entitled "Millimetric Capture of Human Subjects"
11.1.4 Scientific expertise
- Jean-Sébastien Franco was on the recruiting committee (COS) for an assistant professor positions at Grenoble INP - Ensimag, Laboratoire Jean Kuntzmann.
- Sergi Pujades was on the recruiting committee (COS) for an ATER position UFR IM2AG.
- Sergi Pujades was on the recruiting committee (COS) for the Delegations INRIA.
- Sergi Pujades was a reviewer for the RIPEC committe at Université Grenoble Alpes.
11.1.5 Research administration
- Stefanie Wuhrer is référente données for the Inria centre of the Grenoble Alpes University
11.2 Teaching - Supervision - Juries - Educational and pedagogical outreach
- Master: Sergi Pujades, Computer Vision, 54h EqTD, M2R Mosig GVR, Grenoble INP.
- Master: Sergi Pujades, Introduction to Visual Computing, 42h, M1R Mosig GVR, Université Grenoble Alpes.
- Master: Julien Pansiot, Introduction to Visual Computing, 15h EqTD, M1 MoSig, Université Grenoble Alpes.
- Master: Jean-Sébastien Franco, Introduction to Computer Vision, 45h, Ensimag 3rd year, Grenoble INP.
11.2.1 Supervision
- Ph.D. ongoing: Vicente Estopier Castillo. INTENSIVE : Regional Lung Function Inference from Upper Body Surface Motion for Personalized Ventilation Protocols in Acute Respiratory Failure Patients, since 01.10.2023. Supervised by Jean-Sébastien Franco, Sergi Pujades and Sam Bayat.
- Ph.D. ongoing: Gabriel Ravelomanana. Automated medical image segmentation for the foot of patients with rheumatoid arthritis, since 01.11.2024. Supervised by Julien Pansiot and Sergi PUJADES.
- Ph.D. ongoing: Quentin Zoppis, since 01.10.2024. Supervised by François Berard, Laurence Nigay and Sergi Pujades.
- Ph.D. defended: Briac Toussaint, High precision alignment of non-rigid surfaces for 3D performance capture, since 01.10.2021, defended on 16.09.2025, supervised by Jean-Sébastien Franco.
- Ph.D. defended: Rim Rekik. Generation and evaluation of 3D human motion, since 1.11.2021. Defended on 23.06.2025. Supervised by Anne-Hélène Olivier and Stefanie Wuhrer.
- Ph.D. defended: Aymen Merrouche. Learning Non-Rigid 3D Surface Matching, since 1.10.2021, defended 13.11.2025. Supervised by Edmond Boyer and Stefanie Wuhrer.
- Ph.D. ongoing: Antoine Dumoulin. Video-based dynamic garment representation and synthesis, since 1.11.2023. Supervised by Pierre Hellier, Adnane Boukhayma, and Stefanie Wuhrer.
- Ph.D. ongoing: Laura Neschen. Joint 4D reconstruction and correspondence computation, since 1.10.2024. Supervised by Jean-Sébastien Franco and Stefanie Wuhrer.
- Ph.D. ongoing: Samara Ghrer. Analysis of 3D plant growth, since 1.11.2024. Supervised by Christophe Godin, Franck Hétroy-Wheeler, and Stefanie Wuhrer.
- Ph.D. ongoing: Youssef Ben Cheikh. Hair Modeling from IMages, since 01.05.2024. Supervised by Jean-Sébastien Franco.
- Ph.D. ongoing: Felix Küper. "Towards automatic and interpretable General Movement Assessment for early infant diagnosis", since 01.10.2025. Supervised by Sergi Pujades and Nikolas Hesse.
11.2.2 Juries
- Jean-Sébastien Franco presided the Ph.D thesis of Juliette Marrie, Université de Grenoble-Alpes.
- Jean-Sébastien Franco was an invited jury member for the Ph.D thesis of Théo Cheynel, Ecole Polytechnique.
- Jean-Sébastien Franco was jury member for the Ph.D thesis of Aymen Merrouche, Université de Grenoble-Alpes.
- Jean-Sébastien Franco reviewed for the Ph.D thesis of Shubhendu Jena, Université de Rennes.
- Jean-Sébastien Franco was jury member for the HdR defense of Sergi Pujades, Université de Grenoble-Alpes.
- Stefanie Wuhrer reviewed the Ph.D. thesis of Thomas Besnier, Université de Lille.
- Stefanie Wuhrer reviewed the Ph.D. thesis of Théo Cheynel, Institut Polytechinque de Paris.
- Sergi Pujades was jury member for the PhD thesis of Vaëa Tesan, Université de Grenoble-Alpes.
- Sergi Pujades reviewed the Ph.D. thesis of Kilian Chandeleon, Université Clermont Auvergne.
- Sergi Pujades reviewed the Ph.D. thesis of Martin Bertsch, ETH Zurich.
- Sergi Pujades was a permanent jury of the Mosig M2 final year internship defences in September.
11.2.3 Educational and pedagogical outreach
- Sergi Pujades participated in the Thematic day for high-school students on the topic of "The human and the machine", organized at INRIA.
11.3 Popularization
11.3.1 Productions (articles, videos, podcasts, serious games, ...)
- Stefanie Wuhrer participated in the creation of short film Reconstruire le mouvement humain explaining some key results around human reconstruction, generation and retargeting. This film, made in collaboration with Esprit Sorcier TV and targeting children and adolescents, is available at this location.
12 Scientific production
12.1 Major publications
- 1 articleQuality assessment of 3D human animation: Subjective and objective evaluation.IEEE Transactions on Visualization and Computer GraphicsMay 2025, 1-12HALDOI
- 2 inproceedingsProbeSDF: Light Field Probes For Neural Surface Reconstruction.CVPR 2025 - IEEE/CVF Conference on Computer Vision and Pattern RecognitionNashville (Tenessee), United StatesIEEEJune 2025, 1-27HALDOI
- 3 inproceedingsEfficient Finger Model and Accurate Tracking for Hover-and-Touch, Mid-air and Microgesture Interaction.UIST 2025 - 38th Annual ACM Symposium on User Interface Software and TechnologyBusan, South KoreaACMOctober 2025, 1-14HALDOI
12.2 Publications of the year
International journals
International peer-reviewed conferences
Doctoral dissertations and habilitation theses
Reports & preprints