Jonathan Prexl
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Jonathan Prexl has a total of 14 co-authors. Here are the top co-authors by number of shared publications:
- Michael Schmitt
Department of Aerospace Engineering, University of the Bundeswehr Munich, Neubiberg, Germany
4 shared publications - Caroline Margaux Gevaert
University of Twente
2 shared publications - Devis Tuia
École Polytechnique Fédérale de Lausanne (EPFL)
2 shared publications - Jefersson Alex dos Santos
University of Sheffield
2 shared publications - Keiller Nogueira
University of Stirling
2 shared publications - Marc Rußwurm
Wageningen University
2 shared publications
Latest Publications
- 1 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Magazine
- Vol. 13 (3) , 512-532
Better, Not Just More: Data-centric machine learning for Earth observation
Presents corrections to the paper, Better, Not Just More: Data-Centric Machine Learning for Earth Observation.
- 1 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Magazine
- Vol. 13 (3) , 512-532
Better, Not Just More: Data-centric machine learning for Earth observation
Presents corrections to the paper, Better, Not Just More: Data-Centric Machine Learning for Earth Observation.
- 24 April 2025
- Published by Springer Nature
- p. 317-331
SenPa-MAE: Sensor Parameter Aware Masked Autoencoder for Multi-satellite Self-supervised Pretraining
- 24 April 2025
- Published by Springer Nature
- p. 317-331
SenPa-MAE: Sensor Parameter Aware Masked Autoencoder for Multi-satellite Self-supervised Pretraining
- Open Access
SARFormer -- An Acquisition Parameter Aware Vision Transformer for Synthetic Aperture Radar Data
This manuscript introduces SARFormer, a modified Vision Transformer (ViT) architecture designed for processing one or multiple synthetic aperture radar (SAR) images. Given the complex image geometry of SAR data, we propose an acquisition parameter encoding module that significantly guides the learning process, especially in the case of multiple images, leading to improved performance on downstream tasks. We further explore self-supervised pre-training, conduct experiments with limited labeled data, and benchmark our contribution and adaptations thoroughly in ablation experiments against a baseline, where the model is tested on tasks such as height reconstruction and segmentation. Our approach achieves up to 17% improvement in terms of RMSE over baseline models
Open AccessSARFormer -- An Acquisition Parameter Aware Vision Transformer for Synthetic Aperture Radar Data
This manuscript introduces SARFormer, a modified Vision Transformer (ViT) architecture designed for processing one or multiple synthetic aperture radar (SAR) images. Given the complex image geometry of SAR data, we propose an acquisition parameter encoding module that significantly guides the learning process, especially in the case of multiple images, leading to improved performance on downstream tasks. We further explore self-supervised pre-training, conduct experiments with limited labeled data, and benchmark our contribution and adaptations thoroughly in ablation experiments against a baseline, where the model is tested on tasks such as height reconstruction and segmentation. Our approach achieves up to 17% improvement in terms of RMSE over baseline models
- 1 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Magazine
- Vol. 12 (4) , 335-355
Better, Not Just More: Data-centric machine learning for Earth observation
Recent developments and research in modern machine learning have led to substantial improvements in the geospatial field. Although numerous deep learning architectures and models have been proposed, the majority of them have been solely developed on benchmark datasets that lack strong real-world relevance. Furthermore, the performance of many methods has already saturated on these datasets. We argue that a shift from a model-centric view to a complementary data-centric perspective is necessary for further improvements in accuracy, generalization ability, and real impact on enduser applications. Furthermore, considering the entire machine learning cycle—from problem definition to model deployment with feedback—is crucial for enhancing machine learning models that can be reliable in unforeseen situations. This work presents a definition as well as a precise categorization and overview of automated data-centric learning approaches for geospatial data. It highlights the complementary role of data-centric learning with respect to model-centric in the larger machine learning deployment cycle. We review papers across the entire geospatial field and categorize them into different groups. A set of representative experiments shows concrete implementation examples. These examples provide concrete steps to act on geospatial data with data-centric machine learning approaches.
- 1 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Magazine
- Vol. 12 (4) , 335-355
Better, Not Just More: Data-centric machine learning for Earth observation
Recent developments and research in modern machine learning have led to substantial improvements in the geospatial field. Although numerous deep learning architectures and models have been proposed, the majority of them have been solely developed on benchmark datasets that lack strong real-world relevance. Furthermore, the performance of many methods has already saturated on these datasets. We argue that a shift from a model-centric view to a complementary data-centric perspective is necessary for further improvements in accuracy, generalization ability, and real impact on enduser applications. Furthermore, considering the entire machine learning cycle—from problem definition to model deployment with feedback—is crucial for enhancing machine learning models that can be reliable in unforeseen situations. This work presents a definition as well as a precise categorization and overview of automated data-centric learning approaches for geospatial data. It highlights the complementary role of data-centric learning with respect to model-centric in the larger machine learning deployment cycle. We review papers across the entire geospatial field and categorize them into different groups. A set of representative experiments shows concrete implementation examples. These examples provide concrete steps to act on geospatial data with data-centric machine learning approaches.
- Open Access
- 4 November 2024
- Published by Copernicus GmbH in ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
- Vol. X-3-2024, 231-236
Multiscale assessment of land use efficiency in functional urban areas of Munich and Augsburg
This study aims to understand the spatio-temporal variations in land use efficiency (LUE) estimations at different scales of assessment, focusing on the per-capita built-up area within the Functional Urban Areas (FUAs) of Munich and Augsburg, Germany. The analysis is done for the time epochs 1985, 1995, 2005, 2015 and 2020. FUAs are defined as the economic area of influence in each urban area, making the study of land consumption in FUAs crucial for spatial planning and policy analysis. Land use efficiency is computed using built up surface and population data extracted from the Global Human Settlement Layer. This is done at municipal level and 100 m grid level. Furthermore, the effects of using high resolution building footprints for LUE computations are assessed. The analysis shows that urban cores and the adjacent regions are densifying with lower per capita built surface consumption while smaller urban areas utilise more per capita built surface leading to inefficent land use. It is also observed that higher resolution assessments helps in identifying varying intra-regional LUE estimates for a focused urban planning strategy development. Our study concludes that land use efficiency has generally decreased over time in the commuting zones of FUAs while the urban cores have remained efficient.
- 4 November 2024
- Published by Copernicus GmbH in ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
- Vol. X-3-2024, 231-236
Open AccessMultiscale assessment of land use efficiency in functional urban areas of Munich and Augsburg
This study aims to understand the spatio-temporal variations in land use efficiency (LUE) estimations at different scales of assessment, focusing on the per-capita built-up area within the Functional Urban Areas (FUAs) of Munich and Augsburg, Germany. The analysis is done for the time epochs 1985, 1995, 2005, 2015 and 2020. FUAs are defined as the economic area of influence in each urban area, making the study of land consumption in FUAs crucial for spatial planning and policy analysis. Land use efficiency is computed using built up surface and population data extracted from the Global Human Settlement Layer. This is done at municipal level and 100 m grid level. Furthermore, the effects of using high resolution building footprints for LUE computations are assessed. The analysis shows that urban cores and the adjacent regions are densifying with lower per capita built surface consumption while smaller urban areas utilise more per capita built surface leading to inefficent land use. It is also observed that higher resolution assessments helps in identifying varying intra-regional LUE estimates for a focused urban planning strategy development. Our study concludes that land use efficiency has generally decreased over time in the commuting zones of FUAs while the urban cores have remained efficient.
- Open Access
SenPa-MAE: Sensor Parameter Aware Masked Autoencoder for Multi-Satellite Self-Supervised Pretraining
This paper introduces SenPa-MAE, a transformer architecture that encodes the sensor parameters of an observed multispectral signal into the image embeddings. SenPa-MAE can be pre-trained on imagery of different satellites with non-matching spectral or geometrical sensor characteristics. To incorporate sensor parameters, we propose a versatile sensor parameter encoding module as well as a data augmentation strategy for the diversification of the pre-training dataset. This enables the model to effectively differentiate between various sensors and gain an understanding of sensor parameters and the correlation to the observed signal. Given the rising number of Earth observation satellite missions and the diversity in their sensor specifications, our approach paves the way towards a sensor-independent Earth observation foundation model. This opens up possibilities such as cross-sensor training and sensor-independent inference.
Open AccessSenPa-MAE: Sensor Parameter Aware Masked Autoencoder for Multi-Satellite Self-Supervised Pretraining
This paper introduces SenPa-MAE, a transformer architecture that encodes the sensor parameters of an observed multispectral signal into the image embeddings. SenPa-MAE can be pre-trained on imagery of different satellites with non-matching spectral or geometrical sensor characteristics. To incorporate sensor parameters, we propose a versatile sensor parameter encoding module as well as a data augmentation strategy for the diversification of the pre-training dataset. This enables the model to effectively differentiate between various sensors and gain an understanding of sensor parameters and the correlation to the observed signal. Given the rising number of Earth observation satellite missions and the diversity in their sensor specifications, our approach paves the way towards a sensor-independent Earth observation foundation model. This opens up possibilities such as cross-sensor training and sensor-independent inference.