About
About
Embedded Files
News
Poster board assignment: REO workshop papers are presented an boards 58-93 (UCPH: In the main hallway). Please see the poster board assignment behind each paper title below.
Poster session 1: odd poster board ids present
Poster session 2: even poster board ids present
Location: The workshop will run from 9:00 AM to 5:00 PM. The workshop will be taking place at the University of Copenhagen South Campus (Njalsgade 76), at Auditorium (4A.0.69) on the ground floor. Please note that this is not the same location as the main EurIPS conference! It should however be close by via Metro - the closest station being Island Brygge.
Poster format: The poster boards support A0 portrait or A1 landscape.
Please register as soon as possible, following the instructions on eurips.cc:
Note for paper presenters: If you have an accepted workshop paper to present, registration is still available for you. Please contact us on contact@eurips.cc with documentation of paper acceptance attached, and we will do our best to help you out.
The workshop dates are announced: The REO workshop will take place on December 7th, 2025
For 4-page extended abstracts, a short supplementary material is ok (see submission guidelines below!)
Submission deadline is extended to October 20, AoE!
The CMT submission page is online: https://cmt3.research.microsoft.com/REO2025/
About the REO workshop: The Representation Learning for Earth Observation (REO) workshop brings together researchers and practitioners from machine learning, computer vision, and Earth sciences to advance the development of robust, interpretable, and scalable models for monitoring our planet. With the growing availability of large-scale, multimodal Earth observation (EO) data and the rise of general-purpose foundation models, new opportunities and challenges emerge for integrating data-driven approaches across sensing modalities and application domains. REO will provide a forum for presenting novel technical methods, scientific applications, and system-level innovations, fostering cross-disciplinary exchange and collaborations between academia, industry, and policy stakeholders.
Due to EO's huge potential to tackle pressing societal challenges, there has also been a growing interest from the machine learning and computer vision community in recent years. The development of representation learning approaches within the EO domain has gained interest and momentum beyond academia. Notable recent examples include Google DeepMind’s AlphaEarth, IBM-ESA’s Terramind, AllenAI’s Earth System, and Meta’s DINOv3.
This growing interest calls for increased community exchange around the development, deployment, and the practical use of these models. This workshop aims to discuss the following :
Where are we, and how can we move forward as a community?
What are the open challenges in learning representations of EO data?
In the era of general-purpose one-for-all models, what is the role of specialized approaches?
About the EurIPS conference: EurIPS is a European conference officially endorsed by NeurIPS, the most prestigious AI conferences globally; EurIPS showcases cutting-edge research papers that shape the future of artificial intelligence; EurIPS workshops are independent of NeurIPS workshops; the ELLIS UnConference is the kick-off event of EurIPS and welcomes all participants to join.
More information on the EurIPS website: eurips.cc
Speakers
Speakers
Accepted Papers
Accepted Papers
A decade of sea ice concentration retrieved from Sentinel-1 [poster] [poster #58]
A self-supervised multi-source framework and architecture for generative cross-sensor harmonization. [paper, poster] [poster #59]
A Vertical Vegetation Structure Model of Europe [paper, poster] [poster #60]
Altimeter and velocity data fusion for enhanced spatiotemporal resolution of the Ice Sheet elevation [paper, poster] [poster #61]
Beyond Building Footprints: Probing DINOv3 to Map Roof Material and Geometry [paper, poster] [poster #62]
Collaborative Unpaired Multimodal Representation Learning for Satellite Imagery [poster] [poster #63]
Cryo2S1: Cross-Sensor Representation Learning for Sea Ice Radar Freeboard and Leads in Sentinel-1 SAR [paper, poster] [poster #64]
EoS-FM: Can an Ensemble of Specialist Models act as a Generalist Feature Extractor ? [paper, poster] [poster #65]
Gold Exploration using Representations from a Multispectral Autoencoder [paper, poster] [poster #66]
GOMAA-Geo: Goal Modality Agnostic Active Geo-localization [poster] [poster #67]
Harnessing Multi-Modal Co-learning for Missing Earth Observation Modalities [paper, poster] [poster #68]
HELM: Hierarchical and Explicit Label Modeling with Graph Learning for Multi-Label Image Classification [paper] [poster #69]
Latent Field Reduction of Earth Observation Foundation Model [paper, poster] [poster #70]
Learned Image Compression for Earth Observation: Implications for Downstream Segmentation Tasks [paper, poster] [poster #71]
LEPA: Learning Geometric Equivariance in Earth Observation with a Predictive Architecture [paper] [poster #72]
Leveraging Compact Satellite Embeddings and Graph Neural Networks for Large-Scale Poverty Mapping [paper] [poster #73]
M3DRS: Multi-Modal Multispectral Dataset for Remote Sensing [paper, poster] [poster #74]
MAPLE: Multi-Path Adaptive Propagation with Level-Aware Embeddings for Hierarchical Multi-Label Image Classification [paper] [poster #75]
Mixture of Geographical Experts: Disentangling Earth [paper, poster] [poster #76]
NeurEO: dissecting Earth observation embeddings with computational neuroscience [paper] [poster #77]
On-the-Fly OVD Adaptation with FLAME: Few-shot Localization via Active Marginal-Samples Exploration [paper, poster] [poster #78]
Overlap-Free Modality Generalization in Remote Sensing Foundation Models [paper, poster] [poster #79]
SatOSM: Training geospatial foundation models with the Earth’s largest open ground truth [paper, poster] [poster #80]
Scalable Geospatial Data Generation Using AlphaEarth Foundations [poster] [poster #81]
SenForFlood: A Global Dataset for Flood Mapping [poster] [poster #82]
SHRUG-FM: Reliability-Aware Foundation Models for Earth Observation [paper, poster] [poster #83]
SkyCap: Bitemporal VHR Optical–SAR Quartets for Amplitude Change Detection and Foundation-Model Evaluation [paper, poster] [poster #84]
SuperF: Neural Implicit Fields for Multi-Image Super-Resolution [paper] [poster #85]
Towards Methane Detection On Board Satellites [paper] [poster #86]
Towards Large Scale Geostatistical Methane Monitoring with Part-based Object Detection [paper] [poster #87]
Call for Papers
Call for Papers
We invite participants to present their novel work as extended abstracts or discuss recently published work that is relevant for the workshop. All submissions are handled through the CMT platform and will be reviewed by the program committee. Accepted contributions will be presented as posters (format guidelines coming soon), but authors can indicate their interest of giving an oral talk (10min).
Submission via CMT: https://cmt3.research.microsoft.com/REO2025/
Submission guidelines
Novel unpublished work should be formatted using the NeurIPS template with max. 4 pages single column (excluding references, short supplementary material is ok). The submission is double-blind (anonymous authors and reviewers) and non-archival, which means that we will not formally publish the submissions.
Published papers from the last year (after NeurIPS 2024) can be submitted by uploading the formally published paper (single-blind, anonymous reviewers).
Timeline
Submission deadline: October 15 October 20, 2025, AoE (Please register your paper early, this helps us to organize the review process)
Notification of acceptance: October 31, 2025, AoE
Camera ready for extended abstracts: November 28, 2025, AoE
Scope
This EurIPS workshop welcomes contributions spanning technical methods, scientific applications, and system-level innovations across EO, environmental monitoring, and related Earth sciences. Topics of interest include (but are not limited to):
Machine Learning and AI for EO: self-supervised, multimodal, and domain-adaptive models; continual and online learning; foundation models tailored to EO; human-in-the-loop and active learning strategies.
Physics-based and Hybrid Modeling: integration of Radiative Transfer Models (RTMs) and other physical simulators into ML pipelines; hybrid AI–physics models for parameter retrieval and uncertainty quantification.
Ecology and Environmental Monitoring: detection and tracking of land use and land cover change, biodiversity and habitat mapping, phenology, biomass and canopy height estimation, soil and vegetation condition assessment.
Remote Sensing and Satellite Data Processing: handling multimodal sources (e.g., multispectral, SAR, LiDAR, hyperspectral), multi-resolution fusion, temporal change detection, and cross-sensor harmonization.
Embeddings and Compression: learned representations for efficient storage, retrieval, and search in large EO archives; semantic compression for scalable analysis without loss of task-relevant information.
Earth Science Applications: geophysical parameter estimation, urban and rural mapping, monitoring of geohazards such as floods, landslides, or volcanic activity.
Data Curation, Bias, and Accessibility: building globally representative, reproducible, and inclusive EO datasets; mitigating spatial and sensor biases; developing open benchmarks and standardizing EO pipelines.
Technical and Use Case Innovations: novel architectures, training strategies, or processing pipelines with demonstrable impact on real-world EO challenges.
Schedule
Schedule
Location
Location
Bella Center Copenhagen
Bella Center Copenhagen
Center Blvd. 5
Center Blvd. 5
2300 København S
2300 København S
Denmark
Denmark
Organizers and Program Committee
Organizers and Program Committee
Organizers
Loic Landrieu (ENPC)
Begüm Demir (BIFOLD and TU Berlin)
Nico Lang (University of Copenhagen)
Johannes Jakubik (IBM)
Valerio Marsocci (ESA, Φ-lab)
Ruben Cartuyvels (ESA, Φ-lab)
Hui Zhang (University of Copenhagen)
Program Committee
Benedikt Blumenstiel
Dimitri Gominski
Dino Ienco
Gencer Sumbul
Ghjulia Sialelli
Guillaume Astruc
Hui Zhang
Jan Dirk Wegner
Johannes Jakubik
Linus Scheibenreif
Loic Landrieu
Lukas Drees
Marc Rußwurm
Mikolaj Czerkawski
Nico Lang
Nicolas Longepe
Peter Naylor
Riccardo D'Ercole
Ruben Cartuyvels
Sander Jyhne
Steffen Knoblauch
Valerio Marsocci
Venkanna Babu Guthula
Vivien Sainte Fare Garnot
Yohann Perron
Zhitong Xiong
Sponsors
Sponsors
Acknowledgement:
The Microsoft CMT service was used for managing the peer-reviewing process for this conference. This service was provided for free by Microsoft and they bore all expenses, including costs for Azure cloud services as well as for software development and support.
Page updated
Google Sites
Report abuse