AI4EO Spring School 2026 – Vannes, April 8-10

Quick facts

  • Who this is for: Graduate students, junior researchers, and EO enthusiasts
  • Format: 3 days, in-person only, at Université Bretagne Sud, Tohannic Campus, Vannes
  • Topics: Foundation Models for Earth Observation, MLOps, Responsible AI, and Generative Models
  • Methods: Keynote lectures by leading scientists, hands-on coding sessions, and data-driven project with ESA Phi-lab
  • Working Language: English
  • Participation: Upon admission (free of charge)
  • Deadline for Application: 3rd April, 2026

The OBELIX group is co-organizing (together with EJMD Copernicus Master in Digital Earth, Cluster SequoIA – PANORAMIX chair, and ESA Phi-lab) a Spring School on AI4EO, from April 8 to 10, 2026.

The registration is free of charge, but the number of participants is limited. Applications will be reviewed before selection. The application form is available here. We recommend the participants to register as soon as possible.

Academic recognition (3 ECTS) will be provided upon request and successful completion of activities.

The technical programme consists in lectures and hands-on on important topics such as Foundation Models for Earth Observation, MLOps, Responsible AI, and Generative Models. The school will also feature a short data-driven project with ESA Phi-lab, and include some social events.

Preliminary program

Wednesday 8Thursday 9Friday 10
8:30-9:00Registration and welcome (Sebastien Lefèvre)WelcomeWelcome
9:00-10:30Deep learning for change detection in 3D points clouds (Iris de Gélis)Responsible AI – lecture (Pedram Ghamisi)Generative models – lecture (Nicolas Audebert)
10:30-11:00BreakBreakBreak
11:00-12:30Foundation Models – lecture (Stéphane May & Pierre Adorni)Responsible AI – hands-on (Weikang Yu)Generative models – hands-on (Nicolas Audebert)
12:30-14:00LunchLunchLunch
14:00-15:30Foundation Models – hands-on (Stéphane May & Pierre Adorni)Introduction to Deep Learning with TorchGeo – hands-on (Adam Stewart)Data-driven project with ESA Phi-lab
15:30-16:00BreakBreakBreak
16:00-17:30Foundation Models – hands-onMLOps with TorchGeo – hands-on (Adam Stewart)Data-driven project with ESA Phi-lab
17:30-…Social programSocial programClosing

List of speakers

  • Pedram Ghamisi
  • Nicolas Audebert
  • Pierre Adorni is currently a Ph.D. candidate at IRISA, Université Bretagne Sud, France. He received his engineering degree in Computer Science and his M.Sc. in Machine Learning & Optimization from UTC, Compiègne, France in 2024. His research interests include deep learning, computer vision, representation learning, and foundation models for Earth observation. His current work focuses on improving the generalization capabilities of machine learning models across diverse remote sensing datasets and tasks.
  • Adam J. Stewart is a postdoctoral researcher at the Technical University of Munich in the Chair of Data Science in Earth Observation under the guidance of Prof. Xiaoxiang Zhu. His research interests lie at the intersection of machine learning and Earth science, especially remote sensing. He is the creator and lead developer of the popular TorchGeo library (https://github.com/torchgeo/torchgeo), a PyTorch domain library for working with geospatial data and satellite imagery. He received his B.S. from the Department of Earth and Atmospheric Sciences at Cornell University and his Ph.D. from the Department of Computer Science at the University of Illinois Urbana-Champaign.
  • Stéphane May is a research engineer at CNES, the French space agency. He started his career in satellite telecommunications, focusing on payload design. He then transitioned to remote sensing, where he has built over 20 years of expertise in information extraction from satellite data. Through different projects, his technical activities span a wide range, including classification, data fusion, object detection, change detection, Digital Terrain Model extraction, anomaly detection, etc. Throughout his career, he has witnessed and contributed to the evolution from traditional machine learning to advanced deep learning techniques. At CNES, he plays a key role in multiple R&D projects dedicated to enhancing Artificial Intelligence training methodologies. With seven years of experience in edge AI, he significantly contributed to the AI onboard demonstration with the CO3D satellite, launched in 2025. Additionally, he is skilled in operating sensors and remote sensing platforms.
  • Iris de Gélis obtained her M.Eng. in geomatics from the École Nationale des Sciences Géographiques (Marne-la-Vallée, France) in 2018, followed by an M.Sc. in fundamentals of physics for remote sensing from the Institut de Physique du Globe de Paris (IPGP) in 2019. She completed her Ph.D. at Université Bretagne Sud (Vannes, France) in 2023, where her research focused on deep learning methods for change detection in 3D point clouds. Her doctoral work was recognized with the Best PhD Thesis Award from the CNRS research group “Methods and Applications for Geomatics and Spatial Information,” as well as the Second PhD Prize (2023) from the French Association for Pattern Recognition (AFRIF). Since 2023, she has been working as a research engineer at Estellus and is also affiliated with the Paris Observatory (LIRA laboratory). Her current research interests focus on remote sensing using passive microwave data.
  • Weikang Yu is a PhD researcher at the Helmholtz Institute Freiberg (HIF) for Resource Technology, Helmholtz-Zentrum Dresden-Rossendorf (HZDR), Germany. He is also with the Chair of Data Science in Earth Observation, Technical University of Munich (TUM), Germany. He earned his M.Phil and B.E. degrees from the Chinese University of Hong Kong, Shenzhen, and Beihang University, respectively. His research focuses on developing cutting edge deep learning models for remote sensing (RS) applications, as parts of the Artificial Intelligence for Remote Sensing (AI4RS) studies. In particular, he is interested in the studies of change detection, responsible AI, and geospatial foundation models.
  • Sébastien Lefèvre (M.Sc 1999, PhD 2002, Hab. 2009) is a Full Professor in Computer Science at Université Bretagne Sud (UBS) since 2010, promoted to exceptional class by the National Council of Universities in 2023. He is also Adjunct Professor at UiT – The Arctic University of Norway, Tromsø, Norway, since 2023, and Visiting Professor at ESA Phi-lab since 2025. He is involved in numerous activities related to Artificial Intelligence for Earth and Environment Observation, including: founder of the OBELIX team (www.irisa.fr/obelix) within Institute for Research in Computer Science and Random Systems (IRISA), chair of the GeoData Science track of the EMJM Copernicus Master in Digital Earth (www.master-cde.eu), co-founder of the ECML-PKDD MACLEAN workshop series on Machine Learning for Earth Observation, chair of the AI4EO 2025 symposium, coordinator of the UBS-JRC Collaborative Doctoral Program on AI4EO, and holder of the PANORAMIX experienced chair within the SequoIA AI cluster (2025-2029).

How to come to Vannes

  • By train:
    • The TGV (high-speed train) connects Vannes from Paris (Gare Montparnasse) in ~2h 30min (13 connections daily).
    • Regional trains from Rennes and Nantes to Vannes are regularly operated through TER Breizhgo
  • By plane: Rennes or Nantes airport, or Paris CDG airport
    • Nantes Airport: Nantes is well connected to Vannes, with a direct train in ~1h 15min. The airport of Nantes has a bus shuttle to the train station every 20 min.
    • Carpooling with BlaBlaCar can also be a viable option, and the journey takes around 1h. 
    • Paris CDG Airport: there are a few direct TGV connections to Rennes, otherwise through Paris Montparnasse (see train options above)
  • Local transport: The campus is served by the Kicéo bus network (Line 2 and 6, stops at PIBS2 or Tohannic). Vannes is a medium-size city, the campus is reachable from the center with a 30min walk.

Suggestion for accommodation

  • Appart’City Confort Vannes: Walking distance to the campus
  • Numerous hotel and private housing options in Vannes, for instance: Quality Hotel Vannes (Modern amenities, located near the expressway and bus routes), Hôtel Kyriad Vannes Centre (For those wishing to stay in the historic medieval center, 15-minute bus ride to campus)

Suggestion for food 

The university canteen (6 on the map below, school will be held on building 2) provides lunch between 12 and 2 pm at affordable prices. 

Campus map

General Notes