Post-Doctoral Research Visit F - M Postdoc Learning To Reconstruct Blooming Roses In 3D H/F

INRIA

Post-Doctoral Research Visit F/M Postdoc: Learning to reconstruct blooming roses in 3D
Le descriptif de l'offre ci-dessous est en Anglais
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 Grenoble research center groups together almost 600 people in 27 research teams and 8 research support departments.

Staff is present on three campuses in Grenoble, in close collaboration with other research and higher education institutions (University Grenoble Alpes, CNRS, CEA, INRAE, ...), but also with key economic players in the area.

Inria Grenoble is active in the fields of high-performance computing, verification and embedded systems, modeling of the environment at multiple levels, and data science and artificial intelligence. The center is a top-level scientific institute with an extensive network of international collaborations in Europe and the rest of the world.

Contexte et atouts du poste

The postdoc position is part of a joint project between the laboratory Reproduction et Développement des Plants at Ecole Normale Supérieure at Lyon (https://www.ens-lyon.fr/RDP/), Mosaic team at Inria Lyon (https://team.inria.fr/mosaic), and Morpheo team at Inria Grenoble(https://team.inria.fr/morpheo). It is financed by an Inrae Explor'ae funding.

About
The Postdoc will start 01.09.2026 for a duration of 18 months, and be supervised by Stefanie Wuhrer (https://swuhrer.gitlabpages.inria.fr/website/) (Inria Grenoble) and Christophe Godin (https://team.inria.fr/mosaic/welcome/team-members/christophe-godin) (Inria Lyon) for the mathematical and computational aspects as well as by Mohammed Bendahmane (https://www.ens-lyon.fr/RDP/Morphogenese-florale/) (Inrae Lyon) for the biology. It will also be conducted in close collaboration with Julien Pansiot (Inria Grenoble).

Location
The Postdoc will take place at Inria Grenoble with planned regular research visits in Lyon.

Mission confiée

The number of petals of roses varies from 5 to about 300, and the petals of roses with different petal numbers vary in shape and size. Accurately quantifying the size and shape of petals within a rose, at different growth stages, allows to study fundamental biological questions related to living forms morphogenesis. Reconstructing roses in 3D from incomplete and noisy scans in a non-invasive way is critical for this study. This problem is challenging as petals are thin structures and as there are strong occlusions. The 3D reconstruction of flowers has received relatively little study. An early model reconstructs static flowers using botanical priors [1] by building a shape space for flower petals, while not being applicable to growing flowers. Existing methods to reconstruct growing flower petals from vision sensors fit a pre-defined template to observations [2]. This requires interactively defining atemplate per flower, and is not suited to multi-layered petals, as in case of a rose.

This postdoc position is concerned with a data-driven approach that learns to reconstruct open rose flowers in 3D. The key idea is to learn two neural networks that operate on different scales. The first network operates on the scale of the full flower to identify the flower architecture. Our recent work on inferring parametric information for Chenopodium plants is a starting point for this work [3]. The second network operates on a petal scale, to learn information on the shape and position of rose petals, inspired by recent work on leaves [4]. This strategy, combined with explicit contact modeling inspired by works on garment modeling [5] will be explored for the reconstruction of 3D roses. The roses to be studied are modulated on the molecular level to exhibit different numbers of petals while sharing the same genotype by one of the supervisors [6].

References:
1. C. Zhang et al., Data-driven Flower Petal Modeling with Botany Priors, CVPR 2014.
2. Q. Zheng et al., 4D Reconstruction of Blooming Flowers, CGF 2017.
3. S. Ghrer et al., Learning to Infer Parameterized Representations of Plants from 3D Scans, CVPR 2026.
4. Y. Yang et al. NeuraLeaf: Neural Parametric Leaf Models with Shape and Deformation Disentanglement. ICCV 2025.
5. A. Grigorev et al., ContourCraft: Learning to Resolve Intersections in Neural Multi-Garment Simulations, SIGGRAPH 2024.
6. L. François et al., A miR172 target-deficient AP2-like gene correlates with the double flower phenotype in roses, Scientific Reports 2018.

Principales activités

- Studying literature
- Scanning rose flowers at different growth stages
- Implementing and testing neural shape models
- Writing research reports

Compétences

Candidate profiles

Ph.D. in Computer Science or Applied Mathematics.
Solid programming skills, e.g. python and/or C++.
Solid experience with deep learning and shape modeling.
Experience with 3D scanning is a plus.
Experience with plant modelling is a plus.
Good English level. French is not required.

Avantages

- Subsidized meals
- Partial reimbursement of public transport costs
- Leave: 7 weeks of annual leave + 10 extra days off due to RTT (statutory reduction in working hours) + possibility of exceptional leave (sick children, moving home, etc.)
- Possibility of teleworking (90 days / year) and flexible organization of working hours
- Professional equipment available (videoconferencing, loan of computer equipment, etc.)
- Social, cultural and sports events and activities
- Access to vocational training
- Complementary health insuranceunder conditions

Rémunération

2788€ gross salary / month