Offre en lien avec l’Action/le Réseau : – — –/Doctorants

Laboratoire/Entreprise : Image et Vision Artificielle (ImViA)
Durée : 18 mois (might be ex
Contact : alamin.mansouri@u-bourgogne.fr
Date limite de publication : 2024-07-15

Contexte :
This recruitment falls in the framework of the Horizon Europe Research & Innovation Action Entitled CHEMINOVA

https://cordis.europa.eu/project/id/101132442

Partners: UB (France), UVEG (Spain), ICCROM (Italy), UNIPA (Italy, NCA-SSK (Ukraine), CNR-ISAC (Italy), LUH (Germany), 4D-IT (Austria), SKB (Austria), DIADRASIS (Greece), UTC (Romania), ARTCO (Germany).

Connext and scope of Cheminova:

In the 1960s, a chemical experimentation game called CHEMINOVA gained widespread popularity and inspired many individuals to pursue careers in science. The game came with all the essential tools to create a home laboratory, including test tubes, a burner for heating mixtures, clamps to hold tubes, and chemical products. Named after and inspired by this game, the Horizon Europe Research&Innovation Action project entitled ChemiNova focuses on simplifying chemistry and imaging analysis and fostering collaborative conservation research while prioritizing humans as the central focus for technological innovations.

Therefore, ChemiNova project aims to develop an intelligent computational system that goes beyond current technologies to improve the conservation, analysis and monitoring of European Cultural Heritage assets. Using a myriad of data, we will tackle structural and chemical damages, focusing on two specific human-induced threats: climate change and civil conflicts. Inspired by the simplicity of a children’s game, our value relies on the conviction that we will facilitate conservators’ work to the extent that using a single framework they can document, digitise, classify, and share information for CH conservation. Cheminova will organize collaborative acquisition sessions at the pilots: As part of the project in-person technical meetings we have planned collaborative acquisition sessions involving gathering different types of data by ChemiNova partners. This will include hyperspectral images (UVEG), thermal images (CNR-ISAC), RGB images with a camera mounted on UAVs (4D-IT), Appearance attributes from Reflectance Transformation Imaging-RTI (UB) and RGB images with off-the-shelf cameras (UNIPA), among others. Therefore, the technical meetings will mostly take place in the cities related to pilots (Valencia, Palermo, Vienna and Kyiv -if possible).
Furthermore, our impact lies in the fact that we will not build an ad hoc device, but our technology is adapted so that anyone can access it from anywhere. We will involve local communities (citizens) in conservation practices, from providing data (citizen science) to raising awareness on the effects of climate change, natural and human hazards affecting CH.

Sujet :
Involvement of the candidate in the project

The involvement of the successful candidate will be on the entire project with but significant contributions are expected on WP3 and WP4. These contributions will revolve around a) Developing new methods and tools considering multi-dimensional representations of the object, supporting not only spatial information, but also spectral, RTI, semantic and temporal. These 3D models are called enriched 3D models (e3D) as a result of WP3–ChemiModel); b) Developing advanced analysis based on deep learning to automatically detect types of damages on CH assets due to climate change and other human-induced threats (WP4–ChemiAI). c) Processing RTI data acquired by a drone-based system designed for inaccessible zones d) Participating to the management of the project title
a) Building ChemiModel: Design radiometric calibration protocols and methods and apply them to the data; Co-register data from multiple platform positions in a common, metric reference frame (orientation); Reconstruct the object shape as a watertight surface mesh with radiometric information; Refine and enrich the model locally by maps computed from RTI data and simulate visual appearance of objects through interactive relighting; Integrate other models (different epochs, different sensors) and detect changes by differencing; To compound and upload to the ChemiNova database the e3D models with shape and texture(s).
b) Advanced Analysis and Deep Learning (ChemiAI): develop classification techniques for differentiating types of damage of artefacts, buildings and monuments based on the e3D models derived in WP 3 and using methods for deep learning; To develop a methodology for analysing hyperspectral and RTI data in order to obtain information about damage; Integrate the classification methods for e3D models with RGB textures and additional sensor data for an improved prediction of damage; Derive information about damage for the datasets acquired in the pilots so that it can be integrated into the ChemiNova database.
c) Drone-based RTI system building: The generated 3D point cloud may suffer locally from imperfect or low confidence reconstruction (large objects or for zones out of the reach). In such areas, the RTI technique will be used to improve the generated e3D models. A drone-based RTI system will be built in collaboration with 4D-IT. The candidate will be involved in data processing further to acquisition by drone (alignment, stitching, radiometric calibration. Relighting).
d) Management of the project: Strong involvement in the project is expected: attending consortium meetings, preparing and participating to acquisition campaigns, preparing deliverable and monitoring the progress, representing the project and UB in events and conferences.

Profil du candidat :
The candidate must hold a PhD in Computer/computer Science with competences on Deeplearning.
Knowledge on RTI imaging or more globally Appearance imaging are appreciated

Formation et compétences requises :

Adresse d’emploi :
Laboratoire ImViA – Dijon

Document attaché : 202405101448_PostDoc Proposal_FV.pdf