Symposium of the International Future Lab AI4EO 2023

After our successful AI4EO Symposium 2022 we invite you to the Symposium of the International Future Lab AI4EO 2023 October 9th-10th at the TUM Campus in Ottobrunn. Stay tuned to get information about how to sign up for two days full of Research highlights, Poster Sessions, Scientific talks and panel discussions … and of course social events.


Monday, 9 Oct 2023 to 10 Oct 2023 (All day)


onsite, Campus Ottobrunn, Lise-Meitner-Str.9, 85521 Ottobrunn

What to expect:

Organized by the International Future Lab AI4EO and sponsored by the BMBF, the AI4EO Symposium 2023 aims to promote Artificial Intelligence for Earth Observation by addressing cutting-edge AI4EO methods and societally relevant applications.

This symposium will feature keynote talk, as well as oral and poster sessions and ample opportunities for networking and discussion. This symposium will offer an excellent opportunity for participants to share their latest research, collaborate, and learn about recent developments in Data Science in Earth Observation.

Topics (not limited to):

· Reasoning

· Uncertainty quantification

· Ethics in AI4EO

· Green AI

· Generalizability and Transferrability

· Physics-Aware Machine Learning

· Societal Relevant Applications of AI4EO (e.g. urbanization, UN’s SDG, climate science, food security etc)

Important Dates:

Registration deadline: September 30, 2023

Have a sneak peek into our program here:

Abstract submission deadline: September 15, 2023

Acceptance decision: September 30, 2023

P R O G R A M   O C T   9 T H 

09:00    Welcome and Introduction to the Future Lab AI4EO – Prof. Xiaoxiang Zhu

09:30    Oral session I: AI4EO – Reasoning, Uncertainty and Ethics – Chair: Prof. Richard Bamler

4 presentations 12+3 min each

  • Prof. Dr. Jonathan Bamber (TUM AI4EO Future Lab/Uni. Bristol), A statistical framework for spatio-temporal inference: application to global sea level trends
  • Konrad Heidler (TUM), Let’s think it through: Reasoning Strategies in AI for Earth Observation
  • Prof. Dr. Yuanyuan Wang (TUM AI4EO Future Lab), How certain is uncertainty?
  • Prof. Dr. Muhammad Shahzad (TUM AI4EO Future Lab), Towards a Benchmark EO Semantic Segmentation Dataset for Uncertainty Quantification

10.30     coffee break

11:00     Oral session II:  AI4EO for Environment –Chair: Prof. Xiaoxiang Zhu

5 presentations 12+3 min each

  • Dr. Ronny Hänsch (DLR), Towards Cross-Modal Classification of Remote Sensing Imagery
  • Dr. Cornelius Senf (TUM), Mapping forest change across Europe with AI
  • Prof. Dr. Jian Peng (UFZ), Unleashing Machine Learning and Remote Sensing for Land-Atmosphere Interaction
  • Prof. Dr. Pedram Ghamisi (HZDR/IARAI), Managing Limited Labels: Exploring Point-level Labels and Text-to-Image Synthesis
  • Prof. Dr. Guy Schumann (Uni. Bristol), Innovative EO-based flood mapping: the intersection between cloud computing, trustworthy AI and high technology readiness level

12:15     Poster Session I – lightening presentations

12.45     lunch (Group photo after lunch)

14:30     Keynote talk I: Prof. Veronika Eyring, DLR/TUM

Title: Machine learning for improved understanding and projections of climate change

15.30     coffee break/poster viewing

16:00     Oral Session III: AI4EO for Social Good – Chair: Prof. Jonathan Bamber

5 presentations 12+3 min each

  • Dr. Martin Waehlisch (UN), Earth Observation for Sustainable Development and Peacebuilding
  • Dr. Fahad Shah (Nankai University), Exploring land use land cover changes and their effect on urban heat island and land surface temperature: The case of Lahore, Pakistan
  • Yannis Rupp (TU Darmstadt), Towards equation-based modeling of population dynamics in two South Asian Cities
  • Getachew Workineh Gella (Paris Lodron Uni), Dwellings as anomalies: integrating VAEs for unsupervised object counting in humanitarian emergency response
  • Angela Abascal (University of Twente), EO Modelling Environmental Deprivation and Climate Change Impacts in Urban Slums

19:00     Social Event & Dinner

P R O G R A M   O C T   1 0 T H 

09:00    Welcome

09:15     Keynote talk II: Prof. Dr. Hannes Taubenböck, DLR/Uni Würzburg

                Title: Understanding urbanization using EO and AI: Capabilities and Challenges

10:15     coffee break

10:40     Oral session IV: AI4EO – Ethics and Beyond – Chair: Prof. Yuanyuan Wang

5 presentations 12+3 min each

  • Prof. Dr. Mrinalini Kochupillai (TUM AI4EO Future Lab), Ethics in AI4EO
  • Dr. Jiaojiao Tian (DLR), Domain adaptation-from synthetic to real
  • Prof. Dr. Anna Kruspe (TH Nürnberg/TUM), Obtaining geoinformation from social media
  • Prof. Dr. Martin Werner (TUM), Efficient AI for Earth Observation – Towards onboard processing and downlink optimization
  • Dr. Perich Gregor (ETH), EOdal – Open Source Earth Observation for Everyone

12:00     Poster Session II – lightening presentations

12:30     lunch & poster

14:30     Keynote talk III: Dr. Bertrand Le Saux, ESA

Title: Next Generation AI4EO

15:30     Oral Session V: Physics- and Uncertainty-aware AI4EO – Chair: Prof. Muhammad Shahzad

5 presentations 12+3 min each

  • Dr. Piyush Garg (Argonne National Laboratory), Physics-Informed Domain-Aware Convolutional Neural Network Based Atmospheric Radiative Transfer Emulator
  • Qingsong Xu (TUM), Physics-aware Machine Learning: A New Paradigm for Machine Learning and Process-based Hydrology
  • Dr. Alen Turnwald (E:FS TechHub GmbH), Advanced methods for uncertainty quantification and causality in AI applications
  • Björn Tings (DLR), Uncertainty Estimation for Detectability Modelling without Ground Truth
  • Ivica Obadic (TUM), MEDSAT: A Public Health Dataset for England Featuring Medical Prescriptions and Satellite Imagery

16:45     Wrap- up and goodbyes

Detailed list of posters:

Poster Session I: AI4EO applications

Detecting War Destruction in Ukraine using Sentinel-1 Time-Series

Olivier Dietrich, ETH Zurich

A machine-learning-based framework for retrieving water quality parameters in

highly urbanized areas using UAV hyperspectral images

Dr. Bing Liu, Peking University

Comparative Performance Analysis of simple U-Net, Residual Attention U-Net,

and VGG16-U-Net for Inventory InlandWater Bodies

Dr. Mohammadmehdi Saberioon., GFZ

High-resolution satellite images reveal tree growth promoted under urbanization

in South America

Dr. Jianhua Guo, TUM

A Multitask Earth Observation Dataset for Crop Prediction and Climate Applications

Adrian Höhl, TUM

Spatial-temporal dynamics of urban heat island hotspots

Keli Wang, TUM

Towards Large-scale Building Attribute Mapping using Crowdsourced Images:

Scene Text Recognition on Flickr and Problems to be Solved

Dr. Yao Sun, TUM

Geographic Transferability of Building Type Models

Nikolai Skuppin, DLR

Application of deep neural networks to retrieve cloud properties for Sentinel-4

(S4) and TROPOMI / Sentinel-5 Precursor (S5P)

Fabian Romahn, DLR

Estimating Soil Parameters from DESIS Images using Deep Learning

Xiangyu Zhao, TUM

Leveraging High-Resolution Satellite Imagery for Enhanced Organic Farm Inspection

in Compliance with EU Regulations

Tess Posch, Universität für Bodenkultur Wien

Quantifying Tree-species Specific Responses to the Extreme 2022 Drought in


Yixuan Wang; TUM

From Pixels to Species: Empowering Forest Tree Species Mapping with Sentinel-2 Time Series Data Using Deep Learning

Yang Mu, TUM

The spatial-temporal role of urban green space in mitigating land surface temperature

in Chinese mega-cities

Yue Zeng, TUM

Uncovering 22 Years of Deforestation in the Amazon with Artificial Intelligence

Bruno Matosak, Instituto Nacional de Pesquisas Espaciais (INPE)

Cocoa for Gold: mapping the landuse transition between two economically

substantial commodities

Stella Ofori-Ampofo, TUM

Predicting 5-year urban growth probability using satellite data, open data and machine learning

Anna-Lena Erdmann, Cellgrid

Deep-learning-based large-scale forest height estimation with the synergy of ICESAT-2, Sentinel-1 and Sentinel-2 imagery

Dr. Qi Zhang, TUM

A Glance at theWorld Forest by Machine Learning

Haiyin Ye, TUM

Mapping inland aquaculture ponds using deep learning methods

Dr. Qingyu Li, TUM

Poster Session II: AI4EO Methods

Self-Supervised Transfer Learning for Historical Aerial Images

Marvin Burges, TU Wien, Vienna University of Technology

PolyGNN: Polyhedron-based Graph Neural Network for 3D Building Reconstruction from Point Clouds

Zhaiyu Chen, TUM

Generative Adversarial Networks for Long-term Subtle Volcano Deformation Detection

Teo Beker, TUM / DLR

Adaptive Morphology Filter: A Lightweight Module for Deep Hyperspectral Image Classification

Dr. Xin Sun PhD, TUM

SSL4EO-S12: A Large-Scale Multi-Modal, Multi-Temporal Dataset for Self-Supervised Learning in Earth Observation

Yi Wang, TUM / DLR

An inconsistency-aware network for precipitation downscaling

Shan Zhao, TUM

Few-shot Object Detection in Remote Sensing: Lifting the Curse of Incompletely Annotated Novel Objects

Fahong Zhang, TUM

Online Correction of Object Labels for Deep Learning with Incomplete Annotation in Remote Sensing Image Segmentation

Chenying Liu, TUM / DLR

HTC-DC Net: Monocular Height Estimation from Single Remote Sensing Images

Sining Chen, TUM

Hybrid Quantum Deep Learning with Superpixel Encoding for Earth Observation Data Classification

Fan Fan, DLR

Cross-resolution image segmentation for mapping smallest ponds in the Arctic

Tabea Rettelbach, AlfredWegener Institute Helmholtz Centre for Polar and Marine Research

A Comparison of Uncertainty Quantification Methods for Earth Observation Image Regression Data

Nils Lehmann, TUM

Research Project Methane detection and quantification using a multimodal deep learning approach

Enno Tiemann, TUM / OHB Digital Connect GmbH

Glacier Ice Thickness Estimation from Physics-aware Machine Learning

Viola Steidl, TUM

Rapid building damage assessment pipeline near urban flash flood events using unsupervised change detection

Jeremy Eudaric, DLR

Learning high resolution encoding for coarse weather forecasting: Going beyond subgrid eddy forcing

Constantin Le Clei, TUM

Towards AI-based EO Services Using Hyperspectral and Thermal-Infrared Data

Marco Spagnolli, OHB System AG

Simulate LiDAR point cloud for biomass estimation

Dr. Qian Song, TUM

AdaptMatch: Adaptive Matching for Semi-supervised Binary Segmentation of Remote Sensing Images

Wei Huang, TUM

RRSIS: Referring Remote Sensing Image Segmentation

Zhenghang Yuan, TUM