David Völgyes
Research Subjects
David Völgyes has a total of 74 co-authors. Here are the top co-authors by number of shared publications:
- Marius Pedersen
Norwegian University of Science and Technology
4 shared publications Anne Catrine Trægde Martinsen
University of Oslo, Department of Physics
3 shared publicationsPeter Major
Mediso Medical Imaging Systems Ltd., Budapest, Hungary
3 shared publicationsA. Skretting
2 shared publications- Anne Catrine Trægde Martinsen
Department of Physics, University of Oslo, 0316 Oslo, Norway
2 shared publications - Anne H. Schistad Solberg
Department of Informatics, University of Oslo, Oslo, Norway
2 shared publications
Latest Publications
- Open Access
- 25 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 18 (19391404) , 21954-21966
Using Monoscopic Multispectral Earth Observation Images to Predict Terrain Features With Deep Neural Networks
In the field of remote sensing and Earth observation, deep neural networks (DNNs) have established themselves as important tools for many different image analysis applications. Estimation of terrain features from optical satellite imagery is a rarely studied application for which DNNs are well suited because of their ability to extract and combine information at various scales. To predict terrain slopes in optical images, we propose an R2U-Net using global multispectral Sentinel-2 (S2) L2A images as input, and ALOS World 3D DSM elevation data as target data. The R2U-Net takes advantage of a residual unit that benefits deep architecture training, and the recurrent residual convolutional layers provide better feature accumulation. Two models were experimented with; one model trained on only the optical RGB bands and one model trained on all S2 L2A bands. Evaluation of the multispectral- and RGB-trained models showed that the multispectral-trained model performs better than the RGB model, both during training and when evaluated on the test data. The multispectral model performs better overall than the RGB model in all the cases studied. Slope errors typically increase from low- to high-gradient terrain, but not at the same rate as the slope steepness itself, while aspect errors decrease as the models struggle more to predict the slope aspect in low-gradient terrain. This highlights that, in this case, using more spectral bands when predicting terrain slopes helps improve the model predictions. The results have also been shown to depend on the incoming angle of the sunlight, which is mostly due to the topographic shadows that are being cast onto the terrain.
- 25 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 18 (19391404) , 21954-21966
Open AccessUsing Monoscopic Multispectral Earth Observation Images to Predict Terrain Features With Deep Neural Networks
In the field of remote sensing and Earth observation, deep neural networks (DNNs) have established themselves as important tools for many different image analysis applications. Estimation of terrain features from optical satellite imagery is a rarely studied application for which DNNs are well suited because of their ability to extract and combine information at various scales. To predict terrain slopes in optical images, we propose an R2U-Net using global multispectral Sentinel-2 (S2) L2A images as input, and ALOS World 3D DSM elevation data as target data. The R2U-Net takes advantage of a residual unit that benefits deep architecture training, and the recurrent residual convolutional layers provide better feature accumulation. Two models were experimented with; one model trained on only the optical RGB bands and one model trained on all S2 L2A bands. Evaluation of the multispectral- and RGB-trained models showed that the multispectral-trained model performs better than the RGB model, both during training and when evaluated on the test data. The multispectral model performs better overall than the RGB model in all the cases studied. Slope errors typically increase from low- to high-gradient terrain, but not at the same rate as the slope steepness itself, while aspect errors decrease as the models struggle more to predict the slope aspect in low-gradient terrain. This highlights that, in this case, using more spectral bands when predicting terrain slopes helps improve the model predictions. The results have also been shown to depend on the incoming angle of the sunlight, which is mostly due to the topographic shadows that are being cast onto the terrain.
- 5 October 2024
- Published by Springer Nature
- p. 122-131
Introducing Anatomical Constraints in Mitral Annulus Segmentation in Transesophageal Echocardiography
- 5 October 2024
- Published by Springer Nature
- p. 122-131
Introducing Anatomical Constraints in Mitral Annulus Segmentation in Transesophageal Echocardiography
- 19 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 62 (01962892) , 1-10
Terrain-Informed Self-Supervised Learning: Enhancing Building Footprint Extraction From LiDAR Data With Limited Annotations
Estimating building footprint maps from geospatial data is vital in urban planning, development, disaster management, and various other applications. Deep learning methodologies have gained prominence in building segmentation maps, offering the promise of precise footprint extraction without extensive postprocessing. However, these methods face challenges in generalization and label efficiency, particularly in remote sensing, where obtaining accurate labels can be both expensive and time consuming. To address these challenges, we propose terrain-aware self-supervised learning (SSL), tailored to remote sensing, using digital elevation models (DEMs) from light detection and ranging (LiDAR) data. We propose to learn a model to differentiate between bare Earth and superimposed structures enabling the network to implicitly learn domain-relevant features without the need for extensive pixel-level annotations. We test the effectiveness of our approach by evaluating building segmentation performance on test datasets with varying label fractions. Remarkably, with only 1% of the labels (equivalent to 25 labeled examples), our method improves over ImageNet pretraining, showing the advantage of leveraging unlabeled data for feature extraction in the domain of remote sensing. The performance improvement is more pronounced in few-shot scenarios and gradually closes the gap with ImageNet pretraining as the label fraction increases. We test on a dataset characterized by substantial distribution shifts (including resolution variation and labeling errors) to demonstrate the generalizability of our approach. When compared with other baselines, including ImageNet pretraining and more complex architectures, our approach consistently performs better, demonstrating the efficiency and effectiveness of self-supervised terrain-aware feature learning.
- 19 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 62 (01962892) , 1-10
Terrain-Informed Self-Supervised Learning: Enhancing Building Footprint Extraction From LiDAR Data With Limited Annotations
Estimating building footprint maps from geospatial data is vital in urban planning, development, disaster management, and various other applications. Deep learning methodologies have gained prominence in building segmentation maps, offering the promise of precise footprint extraction without extensive postprocessing. However, these methods face challenges in generalization and label efficiency, particularly in remote sensing, where obtaining accurate labels can be both expensive and time consuming. To address these challenges, we propose terrain-aware self-supervised learning (SSL), tailored to remote sensing, using digital elevation models (DEMs) from light detection and ranging (LiDAR) data. We propose to learn a model to differentiate between bare Earth and superimposed structures enabling the network to implicitly learn domain-relevant features without the need for extensive pixel-level annotations. We test the effectiveness of our approach by evaluating building segmentation performance on test datasets with varying label fractions. Remarkably, with only 1% of the labels (equivalent to 25 labeled examples), our method improves over ImageNet pretraining, showing the advantage of leveraging unlabeled data for feature extraction in the domain of remote sensing. The performance improvement is more pronounced in few-shot scenarios and gradually closes the gap with ImageNet pretraining as the label fraction increases. We test on a dataset characterized by substantial distribution shifts (including resolution variation and labeling errors) to demonstrate the generalizability of our approach. When compared with other baselines, including ImageNet pretraining and more complex architectures, our approach consistently performs better, demonstrating the efficiency and effectiveness of self-supervised terrain-aware feature learning.
- Open Access
Terrain-Informed Self-Supervised Learning: Enhancing Building Footprint Extraction from LiDAR Data with Limited Annotations
Estimating building footprint maps from geospatial data is of paramount importance in urban planning, development, disaster management, and various other applications. Deep learning methodologies have gained prominence in building segmentation maps, offering the promise of precise footprint extraction without extensive post-processing. However, these methods face challenges in generalization and label efficiency, particularly in remote sensing, where obtaining accurate labels can be both expensive and time-consuming. To address these challenges, we propose terrain-aware self-supervised learning, tailored to remote sensing, using digital elevation models from LiDAR data. We propose to learn a model to differentiate between bare Earth and superimposed structures enabling the network to implicitly learn domain-relevant features without the need for extensive pixel-level annotations. We test the effectiveness of our approach by evaluating building segmentation performance on test datasets with varying label fractions. Remarkably, with only 1% of the labels (equivalent to 25 labeled examples), our method improves over ImageNet pre-training, showing the advantage of leveraging unlabeled data for feature extraction in the domain of remote sensing. The performance improvement is more pronounced in few-shot scenarios and gradually closes the gap with ImageNet pre-training as the label fraction increases. We test on a dataset characterized by substantial distribution shifts and labeling errors to demonstrate the generalizability of our approach. When compared to other baselines, including ImageNet pretraining and more complex architectures, our approach consistently performs better, demonstrating the efficiency and effectiveness of self-supervised terrain-aware feature learning.
Open AccessTerrain-Informed Self-Supervised Learning: Enhancing Building Footprint Extraction from LiDAR Data with Limited Annotations
Estimating building footprint maps from geospatial data is of paramount importance in urban planning, development, disaster management, and various other applications. Deep learning methodologies have gained prominence in building segmentation maps, offering the promise of precise footprint extraction without extensive post-processing. However, these methods face challenges in generalization and label efficiency, particularly in remote sensing, where obtaining accurate labels can be both expensive and time-consuming. To address these challenges, we propose terrain-aware self-supervised learning, tailored to remote sensing, using digital elevation models from LiDAR data. We propose to learn a model to differentiate between bare Earth and superimposed structures enabling the network to implicitly learn domain-relevant features without the need for extensive pixel-level annotations. We test the effectiveness of our approach by evaluating building segmentation performance on test datasets with varying label fractions. Remarkably, with only 1% of the labels (equivalent to 25 labeled examples), our method improves over ImageNet pre-training, showing the advantage of leveraging unlabeled data for feature extraction in the domain of remote sensing. The performance improvement is more pronounced in few-shot scenarios and gradually closes the gap with ImageNet pre-training as the label fraction increases. We test on a dataset characterized by substantial distribution shifts and labeling errors to demonstrate the generalizability of our approach. When compared to other baselines, including ImageNet pretraining and more complex architectures, our approach consistently performs better, demonstrating the efficiency and effectiveness of self-supervised terrain-aware feature learning.
- Open Access
- 23 January 2023
- Published by UiT The Arctic University of Norway in Proceedings of the Northern Lights Deep Learning Workshop
- Vol. 4
Semi- and weak-supervised learning for Norwegian tree species detection
Tree species mapping of Norwegian production forests is a time-consuming process as forest associations largely rely on manual interpretation of earth observation data. Deep learning based image segmentation techniques have the potential to improve automated tree species classification, but a major challenge is the limited quality and availability of training data. Semi-supervised techniques could alleviate the need for training label and weak supervision enables handling coarse-grained and noisy labels. In this study, we evaluated the added value of semi-supervised deep learning methods in a weakly supervised setting. Specifically, consistency training and pseudo-labeling are applied for tree species classification from aerial ortho imagery in Norway. The techniques are generic and relevant for the wider earth observation domain, especially for other land cover segmentation tasks. The results show that consistency training gives a significant performance increase. Pseudo-labeling on the other hand does not, potentially this is due to varying convergence speeds for different classes causing confirmation bias or a partial violation of the cluster assumption.
- 23 January 2023
- Published by UiT The Arctic University of Norway in Proceedings of the Northern Lights Deep Learning Workshop
- Vol. 4
Open AccessSemi- and weak-supervised learning for Norwegian tree species detection
Tree species mapping of Norwegian production forests is a time-consuming process as forest associations largely rely on manual interpretation of earth observation data. Deep learning based image segmentation techniques have the potential to improve automated tree species classification, but a major challenge is the limited quality and availability of training data. Semi-supervised techniques could alleviate the need for training label and weak supervision enables handling coarse-grained and noisy labels. In this study, we evaluated the added value of semi-supervised deep learning methods in a weakly supervised setting. Specifically, consistency training and pseudo-labeling are applied for tree species classification from aerial ortho imagery in Norway. The techniques are generic and relevant for the wider earth observation domain, especially for other land cover segmentation tasks. The results show that consistency training gives a significant performance increase. Pseudo-labeling on the other hand does not, potentially this is due to varying convergence speeds for different classes causing confirmation bias or a partial violation of the cluster assumption.
- Open Access
- 21 September 2022
- Published by University of Oslo Library in Nordic Machine Intelligence
- Vol. 2 (3)
Our MapAI approach: focusing on data pipeline and loss functions
Precision building detection is a difficult challenge because resolution, lighting conditions, and image quality greatly influence the performance of machine learning models. Additionally, the building types, settlement structure, road structure, soil color and texture, vegetation, and car types can also affect image segmentation, making the solutions local or regional. In this paper, we describe the solution for the MapAI challenge submitted by the ATELIER team. We focused on two primary parts: data processing and loss functions. Our main insights were that the data can be effectively resampled by exploiting its structure and that the boundary intersection over union metric is very forgiving, not giving the network enough incentive to refine the borders. The segmentation was performed with a standard 5-level deep U-Net with an additional conditional random fields (CRF) denoiser.
- 21 September 2022
- Published by University of Oslo Library in Nordic Machine Intelligence
- Vol. 2 (3)
Open AccessOur MapAI approach: focusing on data pipeline and loss functions
Precision building detection is a difficult challenge because resolution, lighting conditions, and image quality greatly influence the performance of machine learning models. Additionally, the building types, settlement structure, road structure, soil color and texture, vegetation, and car types can also affect image segmentation, making the solutions local or regional. In this paper, we describe the solution for the MapAI challenge submitted by the ATELIER team. We focused on two primary parts: data processing and loss functions. Our main insights were that the data can be effectively resampled by exploiting its structure and that the boundary intersection over union metric is very forgiving, not giving the network enough incentive to refine the borders. The segmentation was performed with a standard 5-level deep U-Net with an additional conditional random fields (CRF) denoiser.
- Open Access
- 10 May 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 10, 51472-51486
Mitral Annulus Segmentation and Anatomical Orientation Detection in TEE Images Using Periodic 3D CNN
Segmentation of the mitral annulus is often an important step in cardiac examinations. We propose a robust 3D method for predicting the anatomical orientation and segmentation of the mitral annulus in 3D transesophageal echocardiography. The method takes advantage of the circular anatomy of the annulus by utilizing cylinder coordinate samples and a 3D convolutional neural network with circular convolutions. Furthermore, the paper proposes new landmark detection loss functions based on the earth mover’s distance. The method’s effectiveness was demonstrated by training a HighRes3dNet model and evaluating its performance on a separate test set consisting of 135 frames from 19 examinations. The obtained coordinate prediction error was 1.96±1.62 mm, and the anatomical orientation prediction error was 9.7°±15.8°. The robust and fully automatic mitral annulus segmentation and orientation prediction provided by the method can ease the workload of clinicians and provide time savings in clinics.
- 10 May 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 10, 51472-51486
Open AccessMitral Annulus Segmentation and Anatomical Orientation Detection in TEE Images Using Periodic 3D CNN
Segmentation of the mitral annulus is often an important step in cardiac examinations. We propose a robust 3D method for predicting the anatomical orientation and segmentation of the mitral annulus in 3D transesophageal echocardiography. The method takes advantage of the circular anatomy of the annulus by utilizing cylinder coordinate samples and a 3D convolutional neural network with circular convolutions. Furthermore, the paper proposes new landmark detection loss functions based on the earth mover’s distance. The method’s effectiveness was demonstrated by training a HighRes3dNet model and evaluating its performance on a separate test set consisting of 135 frames from 19 examinations. The obtained coordinate prediction error was 1.96±1.62 mm, and the anatomical orientation prediction error was 9.7°±15.8°. The robust and fully automatic mitral annulus segmentation and orientation prediction provided by the method can ease the workload of clinicians and provide time savings in clinics.
- Open Access
- 25 July 2018
- Published by MDPI AG in Algorithms
- Vol. 11 (8) , 111
A Weighted Histogram-Based Tone Mapping Algorithm for CT Images
Computed Tomography (CT) images have a high dynamic range, which makes visualization challenging. Histogram equalization methods either use spatially invariant weights or limited kernel size due to the complexity of pairwise contribution calculation. We present a weighted histogram equalization-based tone mapping algorithm which utilizes Fast Fourier Transform for distance-dependent contribution calculation and distance-based weights. The weights follow power-law without distance-based cut-off. The resulting images have good local contrast without noticeable artefacts. The results are compared to eight popular tone mapping operators.
- 25 July 2018
- Published by MDPI AG in Algorithms
- Vol. 11 (8) , 111
Open AccessA Weighted Histogram-Based Tone Mapping Algorithm for CT Images
Computed Tomography (CT) images have a high dynamic range, which makes visualization challenging. Histogram equalization methods either use spatially invariant weights or limited kernel size due to the complexity of pairwise contribution calculation. We present a weighted histogram equalization-based tone mapping algorithm which utilizes Fast Fourier Transform for distance-dependent contribution calculation and distance-based weights. The weights follow power-law without distance-based cut-off. The resulting images have good local contrast without noticeable artefacts. The results are compared to eight popular tone mapping operators.
- Open Access
- 24 July 2018
- Published by MDPI AG in Algorithms
- Vol. 11 (8) , 110
Image De-Quantization Using Plate Bending Model
Discretized image signals might have a lower dynamic range than the display. Because of this, false contours might appear when the image has the same pixel value for a larger region and the distance between pixel levels reaches the noticeable difference threshold. There have been several methods aimed at approximating the high bit depth of the original signal. Our method models a region with a bended plate model, which leads to the biharmonic equation. This method addresses several new aspects: the reconstruction of non-continuous regions when foreground objects split the area into separate regions; the incorporation of confidence about pixel levels, making the model tunable; and the method gives a physics-inspired way to handle local maximal/minimal regions. The solution of the biharmonic equation yields a smooth high-order signal approximation and handles the local maxima/minima problems.
- 24 July 2018
- Published by MDPI AG in Algorithms
- Vol. 11 (8) , 110
Open AccessImage De-Quantization Using Plate Bending Model
Discretized image signals might have a lower dynamic range than the display. Because of this, false contours might appear when the image has the same pixel value for a larger region and the distance between pixel levels reaches the noticeable difference threshold. There have been several methods aimed at approximating the high bit depth of the original signal. Our method models a region with a bended plate model, which leads to the biharmonic equation. This method addresses several new aspects: the reconstruction of non-continuous regions when foreground objects split the area into separate regions; the incorporation of confidence about pixel levels, making the model tunable; and the method gives a physics-inspired way to handle local maximal/minimal regions. The solution of the biharmonic equation yields a smooth high-order signal approximation and handles the local maxima/minima problems.
- Open Access
- 1 March 2018
- Published by Elsevier in Journal of Forensic Radiology and Imaging
- Vol. 12, 25-30
Applicability of a clinical cardiac CT protocol in post mortem studies
- 1 March 2018
- Published by Elsevier in Journal of Forensic Radiology and Imaging
- Vol. 12, 25-30
Open AccessApplicability of a clinical cardiac CT protocol in post mortem studies