Karun Dayal
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Research Interests
Karun Dayal has a total of 52 co-authors. Here are the top co-authors by number of shared publications:
- Dino Ienco
INRAE, UMR TETIS, University of Montpellier, Montpellier, France
3 shared publications - Jean-Matthieu Monnet
Université Grenoble Alpes, INRAE, LESSEM, 2 rue de la Papeterie-BP 76, F-38402 St-Martin-d'Hères, France
3 shared publications - Kamel Lahssini
TETIS, INRAE, Montpellier, France
3 shared publications - Sylvie Durrieu
UMR TETIS, Univ Montpellier, AgroParisTech, CIRAD, CNRS, INRAE, Montpellier, France
3 shared publications - Kim Calders
Q-ForestLab, Department of Environment, Faculty of Bioscience Engineering, Ghent University, Ghent, Belgium
2 shared publications - Wout Cherlet
Ghent University
2 shared publications
Latest Publications
- Open Access
- 10 September 2026
- Published by Elsevier in SSRN Electronic Journal
occpy: 3D occlusion mapping of laser scanning point clouds
Laser scanning is a key tool for quantifying three-dimensional structure of the environment. Every acquisition is affected by occlusion, where intervening objects block laser pulses and leave parts of the scene unobserved. Occlusion is a widely recognized limitation of laser scanning, yet it is rarely quantified. We present occpy, an open-source Python package that maps occlusion by tracing laser pulses through 3D space. Given point cloud data and scanner location(s), pulses are reconstructed and traced on a 3D voxel grid, producing a 3D map of filled, empty, unobserved and occluded volume. It supports static and moving laser scanning platforms, and features an efficient C++ ray tracing core, interactive 3D visualization, and easy installation through PyPI. Its open-source and modular design enables the integration of occlusion mapping in standard point cloud processing and analysis workflows. While occpy can be used for any laser scanning acquisition, in this paper we focus on the specific use case of laser scanning in forests.
- 10 September 2026
- Published by Elsevier in SSRN Electronic Journal
Open Accessoccpy: 3D occlusion mapping of laser scanning point clouds
Laser scanning is a key tool for quantifying three-dimensional structure of the environment. Every acquisition is affected by occlusion, where intervening objects block laser pulses and leave parts of the scene unobserved. Occlusion is a widely recognized limitation of laser scanning, yet it is rarely quantified. We present occpy, an open-source Python package that maps occlusion by tracing laser pulses through 3D space. Given point cloud data and scanner location(s), pulses are reconstructed and traced on a 3D voxel grid, producing a 3D map of filled, empty, unobserved and occluded volume. It supports static and moving laser scanning platforms, and features an efficient C++ ray tracing core, interactive 3D visualization, and easy installation through PyPI. Its open-source and modular design enables the integration of occlusion mapping in standard point cloud processing and analysis workflows. While occpy can be used for any laser scanning acquisition, in this paper we focus on the specific use case of laser scanning in forests.
- 7 April 2026
- Published by Agence Bibliographique de l'Enseignement Supérieur
Developing the potential of airborne lidar systems for the sustainable management of forests : Accounting for and managing the impacts of lidar scan angle on ABA model predictions of forest attributes
Développement du potentiel du Lidar aéroporté pour la gestion durable des forêts : prise en compte et gestion des effets de l’angle de balayage sur les prédictions d’attributs forestiers à l’aide de modèles surfaciques (ABA): Information measured by lidar depends on the observed vegetation and the acquisition geometry, which is a function of the acquisition parameters and the terrain properties. The thesis aims to understand the relationship between lidar acquisition geometry and forest attribute predictions, focusing on the assessment and management of impacts of lidar scan angle on lidar metrics and ABA models. Four different forest types were studied with three forest types (broadleaf, coniferous and mixed) in mountainous terrain and one forest type (riparian) in relatively flat terrain . The thesis was divided into three parts. The first part assessed the effect of lidar scan angle on lidar metrics commonly used in ABA predictions. It was observed that different lidar metrics behave differently under changing scan angles. Subsequently, the effect of including metrics with different sensitivities to scan angle was investigated in the second part of the study. A model involving a set of predefined metrics with different sensitivities to scan angle was used. Existing lidar datasets were resampled based on the flight lines 1) to simulate lidar acquisitions with different scan geometries, 2) to build models for a set of scan patterns and 3) to further compare the quality of estimations resulting from each scan pattern. These comparisons highlighted that introducing metrics sensitive to scan angle led to a decrease in model robustness. Also, the variation in the accuracy of ABA models was found to be higher for datasets consisting of point clouds scanned from only one flight line as opposed to those consisting of point clouds scanned from multiple flight lines. The normalisation of lidar metrics sensitive to scan angle was also attempted using voxelisation. Voxel-based metrics contributed by increasing either the precision or the accuracy, or both. In the last part of the study, the terrain properties and acquisition parameters were considered explicitly. As the interaction between lidar acquisition parameters, terrain, and vegetation properties can be complex, neural networks were used to model the relationships between various lidar metrics and the acquisition geometry, resulting in significantly better ABA predictions.
- 7 April 2026
- Published by Agence Bibliographique de l'Enseignement Supérieur
Developing the potential of airborne lidar systems for the sustainable management of forests : Accounting for and managing the impacts of lidar scan angle on ABA model predictions of forest attributes
Développement du potentiel du Lidar aéroporté pour la gestion durable des forêts : prise en compte et gestion des effets de l’angle de balayage sur les prédictions d’attributs forestiers à l’aide de modèles surfaciques (ABA): Information measured by lidar depends on the observed vegetation and the acquisition geometry, which is a function of the acquisition parameters and the terrain properties. The thesis aims to understand the relationship between lidar acquisition geometry and forest attribute predictions, focusing on the assessment and management of impacts of lidar scan angle on lidar metrics and ABA models. Four different forest types were studied with three forest types (broadleaf, coniferous and mixed) in mountainous terrain and one forest type (riparian) in relatively flat terrain . The thesis was divided into three parts. The first part assessed the effect of lidar scan angle on lidar metrics commonly used in ABA predictions. It was observed that different lidar metrics behave differently under changing scan angles. Subsequently, the effect of including metrics with different sensitivities to scan angle was investigated in the second part of the study. A model involving a set of predefined metrics with different sensitivities to scan angle was used. Existing lidar datasets were resampled based on the flight lines 1) to simulate lidar acquisitions with different scan geometries, 2) to build models for a set of scan patterns and 3) to further compare the quality of estimations resulting from each scan pattern. These comparisons highlighted that introducing metrics sensitive to scan angle led to a decrease in model robustness. Also, the variation in the accuracy of ABA models was found to be higher for datasets consisting of point clouds scanned from only one flight line as opposed to those consisting of point clouds scanned from multiple flight lines. The normalisation of lidar metrics sensitive to scan angle was also attempted using voxelisation. Voxel-based metrics contributed by increasing either the precision or the accuracy, or both. In the last part of the study, the terrain properties and acquisition parameters were considered explicitly. As the interaction between lidar acquisition parameters, terrain, and vegetation properties can be complex, neural networks were used to model the relationships between various lidar metrics and the acquisition geometry, resulting in significantly better ABA predictions.
- 3 November 2025
- Published by Elsevier in ISPRS Journal of Photogrammetry and Remote Sensing
- Vol. 231, 230-247
Benchmarking tree instance segmentation of terrestrial laser scanning point clouds
- 3 November 2025
- Published by Elsevier in ISPRS Journal of Photogrammetry and Remote Sensing
- Vol. 231, 230-247
Benchmarking tree instance segmentation of terrestrial laser scanning point clouds
- Open Access
- 12 April 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 16 (19391404) , 3531-3544
Enhancing Forest Attribute Prediction by Considering Terrain and Scan Angles From Lidar Point Clouds: A Neural Network Approach
Sensitivity of lidar metrics to scan angle can affect the robustness of area-based approach (ABA) models, and modelling the interplay of scan geometry and terrain properties can be complex. The study hypothesizes that neural networks can manage the interplay of lidar acquisition parameters, terrain properties, and vegetation characteristics to improve ABA models. The study area is in Massif des Bauges Natural Regional Park, eastern France, comprising 291 field plots in a mountainous environment with broadleaf, coniferous, and mixed forest types. Field plots were scanned with a high overlap from multiple flight lines and the corresponding point clouds were considered independently to expand the standard ABA dataset (291 observations) to create a dataset containing 1095 independent observations. Computation of lidar, terrain, and scan angle metrics for each point cloud associated each observation in the expanded dataset with the scan information in addition to the lidar and terrain information. A multilayer perceptron (MLP) was used to model basal area and total volume to compare the predictions resulting from standard and expanded ABA datasets. With expanded datasets containing lidar, terrain, and scan information, the R 2 for the median predictions per plot were higher (R 2 of 0.83 and 0.85 for BA and V tot ) than predictions with standard datasets (R 2 of 0.66( BA ) and 0.71( V tot )) containing only lidar metrics. It also outperformed an MLP model for a dataset with lidar and terrain information [R 2 of 0.77( BA and V tot )]. The MLP performed better than Random forest regression, which could not sufficiently exploit additional terrain and scan information.
- 12 April 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 16 (19391404) , 3531-3544
Open AccessEnhancing Forest Attribute Prediction by Considering Terrain and Scan Angles From Lidar Point Clouds: A Neural Network Approach
Sensitivity of lidar metrics to scan angle can affect the robustness of area-based approach (ABA) models, and modelling the interplay of scan geometry and terrain properties can be complex. The study hypothesizes that neural networks can manage the interplay of lidar acquisition parameters, terrain properties, and vegetation characteristics to improve ABA models. The study area is in Massif des Bauges Natural Regional Park, eastern France, comprising 291 field plots in a mountainous environment with broadleaf, coniferous, and mixed forest types. Field plots were scanned with a high overlap from multiple flight lines and the corresponding point clouds were considered independently to expand the standard ABA dataset (291 observations) to create a dataset containing 1095 independent observations. Computation of lidar, terrain, and scan angle metrics for each point cloud associated each observation in the expanded dataset with the scan information in addition to the lidar and terrain information. A multilayer perceptron (MLP) was used to model basal area and total volume to compare the predictions resulting from standard and expanded ABA datasets. With expanded datasets containing lidar, terrain, and scan information, the R 2 for the median predictions per plot were higher (R 2 of 0.83 and 0.85 for BA and V tot ) than predictions with standard datasets (R 2 of 0.66( BA ) and 0.71( V tot )) containing only lidar metrics. It also outperformed an MLP model for a dataset with lidar and terrain information [R 2 of 0.77( BA and V tot )]. The MLP performed better than Random forest regression, which could not sufficiently exploit additional terrain and scan information.
- Open Access
- 24 January 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in TechRxiv
Enhancing forest attribute prediction by considering terrain and scan angles from Lidar point clouds: a neural network approach
Sensitivity of lidar metrics to scan angle can affect the robustness of area-based approach (ABA) models, and modelling the interplay of scan geometry and terrain properties can be complex. The study hypothesises that neural networks can manage the interplay of lidar acquisition parameters, terrain properties, and vegetation characteristics to improve ABA models. The study area is in Massif des Bauges Natural Regional Park, eastern France, comprising 291 field plots in a mountainous environment with broadleaf, coniferous, and mixed forest types. Field plots were scanned with a high overlap from multiple flight lines and the corresponding point clouds were considered independently to expand the standard ABA dataset (291 observations) to create a dataset containing 1095 independent observations. Computation of lidar, terrain and scan metrics for each point cloud associated each observation in the expanded dataset with the scan information in addition to the lidar and terrain information. A multilayer perceptron (MLP) was used to model basal area and total volume to compare the predictions resulting from standard and expanded ABA datasets. With expanded datasets containing lidar, terrain and scan information, the R² for the median predictions per plot were higher (R² of 0.83 and 0.85 for BA and V tot ) than predictions with standard datasets(R² of 0.66(BA) and 0.71( V tot )) containing only lidar metrics. It also outperformed an MLP model for a dataset with lidar and terrain information (R² of 0.77 ( BA and V tot )). The MLP performed better than RF regression, which could not sufficiently exploit additional terrain and scan information.
- 24 January 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in TechRxiv
Open AccessEnhancing forest attribute prediction by considering terrain and scan angles from Lidar point clouds: a neural network approach
Sensitivity of lidar metrics to scan angle can affect the robustness of area-based approach (ABA) models, and modelling the interplay of scan geometry and terrain properties can be complex. The study hypothesises that neural networks can manage the interplay of lidar acquisition parameters, terrain properties, and vegetation characteristics to improve ABA models. The study area is in Massif des Bauges Natural Regional Park, eastern France, comprising 291 field plots in a mountainous environment with broadleaf, coniferous, and mixed forest types. Field plots were scanned with a high overlap from multiple flight lines and the corresponding point clouds were considered independently to expand the standard ABA dataset (291 observations) to create a dataset containing 1095 independent observations. Computation of lidar, terrain and scan metrics for each point cloud associated each observation in the expanded dataset with the scan information in addition to the lidar and terrain information. A multilayer perceptron (MLP) was used to model basal area and total volume to compare the predictions resulting from standard and expanded ABA datasets. With expanded datasets containing lidar, terrain and scan information, the R² for the median predictions per plot were higher (R² of 0.83 and 0.85 for BA and V tot ) than predictions with standard datasets(R² of 0.66(BA) and 0.71( V tot )) containing only lidar metrics. It also outperformed an MLP model for a dataset with lidar and terrain information (R² of 0.77 ( BA and V tot )). The MLP performed better than RF regression, which could not sufficiently exploit additional terrain and scan information.
- Open Access
- 19 May 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 15 (19391404) , 4337-4348
Combining LiDAR Metrics and Sentinel-2 Imagery to Estimate Basal Area and Wood Volume in Complex Forest Environment via Neural Networks
Forest ecosystems play a fundamental role in natural balances and climate mechanisms through their contribution to global carbon storage. Their sustainable management and conservation is crucial in the current context of global warming and biodiversity conservation. To tackle such challenges, earth observation data have been identified as a valuable source of information. While earth observation data constitute an unprecedented opportunity to monitor forest ecosystems, its effective exploitation still poses serious challenges since multimodal information needs to be combined to describe complex natural phenomena. To deal with this particular issue in the context of structure and biophysical variables estimation for forest characterization, we propose a new deep learning-based fusion strategy to combine together high density three-dimensional (3-D) point clouds acquired by airborne laser scanning with high-resolution optical imagery. In order to manage and fully exploit the available multimodal information, we implement a two-branch late fusion deep learning architecture taking advantage of the specificity of each modality. On the one hand, a 2-D CNN branch is devoted to the analysis of Sentinel-2 time series data, and on the other hand, a multilayer perceptron branch is dedicated to the processing of LiDAR-derived information. The performance of our framework is evaluated on two forest variables of interest: total volume and basal area at stand level. The obtained results underline that the availability of multimodal remote sensing data is not a direct synonym of performance improvements but, the way in which they are combined together is of paramount importance.
- 19 May 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 15 (19391404) , 4337-4348
Open AccessCombining LiDAR Metrics and Sentinel-2 Imagery to Estimate Basal Area and Wood Volume in Complex Forest Environment via Neural Networks
Forest ecosystems play a fundamental role in natural balances and climate mechanisms through their contribution to global carbon storage. Their sustainable management and conservation is crucial in the current context of global warming and biodiversity conservation. To tackle such challenges, earth observation data have been identified as a valuable source of information. While earth observation data constitute an unprecedented opportunity to monitor forest ecosystems, its effective exploitation still poses serious challenges since multimodal information needs to be combined to describe complex natural phenomena. To deal with this particular issue in the context of structure and biophysical variables estimation for forest characterization, we propose a new deep learning-based fusion strategy to combine together high density three-dimensional (3-D) point clouds acquired by airborne laser scanning with high-resolution optical imagery. In order to manage and fully exploit the available multimodal information, we implement a two-branch late fusion deep learning architecture taking advantage of the specificity of each modality. On the one hand, a 2-D CNN branch is devoted to the analysis of Sentinel-2 time series data, and on the other hand, a multilayer perceptron branch is dedicated to the processing of LiDAR-derived information. The performance of our framework is evaluated on two forest variables of interest: total volume and basal area at stand level. The obtained results underline that the availability of multimodal remote sensing data is not a direct synonym of performance improvements but, the way in which they are combined together is of paramount importance.
- 13 January 2022
- Published by Wiley in Journal of Ecology
- Vol. 110 (2) , 282-300
Unveil the unseen: Using LiDAR to capture time‐lag dynamics in the herbaceous layer of European temperate forests
To understand time‐lag dynamics in the response of biodiversity to contemporary environmental changes (e.g. macroclimate warming and atmospheric pollution), we need to consider former anthropogenic forcing factors such as past land uses and management practices that can have both compounding and confounding effects. This is especially true in European temperate forests, where legacies from past human activities have left strong imprints on today’s understorey plant species composition, generating long‐term lagging effects which can be mistakenly attributed to more recent macro‐environmental changes. By combining the expertise of plant, soil and historical ecologists together with remote sensing scientists, we review the potential of light detection and ranging (LiDAR) to unveil ghosts from the past in terms of former land uses and management practices. We show that imprints from past land uses and management practices can still be captured today through well‐chosen LiDAR‐derived variables describing, at sub‐decimetre scale, the vertical and horizontal micro‐variations of vegetation and terrain structure hidden below treetops. Synthesis. We encourage plant and soil ecologists to use LiDAR data and to work with historians, archaeologists and remote sensing scientists in order to select meaningful LiDAR‐derived variables to account for time‐lagged biotic responses to long‐term macro‐environmental changes.
- 13 January 2022
- Published by Wiley in Journal of Ecology
- Vol. 110 (2) , 282-300
Unveil the unseen: Using LiDAR to capture time‐lag dynamics in the herbaceous layer of European temperate forests
To understand time‐lag dynamics in the response of biodiversity to contemporary environmental changes (e.g. macroclimate warming and atmospheric pollution), we need to consider former anthropogenic forcing factors such as past land uses and management practices that can have both compounding and confounding effects. This is especially true in European temperate forests, where legacies from past human activities have left strong imprints on today’s understorey plant species composition, generating long‐term lagging effects which can be mistakenly attributed to more recent macro‐environmental changes. By combining the expertise of plant, soil and historical ecologists together with remote sensing scientists, we review the potential of light detection and ranging (LiDAR) to unveil ghosts from the past in terms of former land uses and management practices. We show that imprints from past land uses and management practices can still be captured today through well‐chosen LiDAR‐derived variables describing, at sub‐decimetre scale, the vertical and horizontal micro‐variations of vegetation and terrain structure hidden below treetops. Synthesis. We encourage plant and soil ecologists to use LiDAR data and to work with historians, archaeologists and remote sensing scientists in order to select meaningful LiDAR‐derived variables to account for time‐lagged biotic responses to long‐term macro‐environmental changes.
- 1 December 2019
- Published by Elsevier in Digital Applications in Archaeology and Cultural Heritage
- Vol. 15, e00117
Diagnostic utilisation of ground based imaging and non-imaging sensors for digital documentation of heritage sites
- 1 December 2019
- Published by Elsevier in Digital Applications in Archaeology and Cultural Heritage
- Vol. 15, e00117
Diagnostic utilisation of ground based imaging and non-imaging sensors for digital documentation of heritage sites
- Open Access
- 5 July 2017
- Published by Copernicus GmbH in The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
- Vol. XLII-4/W2, 67-71
COMPARATIVE ANALYSIS OF 3D POINT CLOUDS GENERATED FROM A FREEWARE AND TERRESTRIAL LASER SCANNER
Keywords: Laser Scanning, Close Range Photogrammetry, Point clouds, Heritage, VSfM In the recent past, several heritage structures have faced destruction due to both human-made incidents and natural calamities that have caused a great loss to the human race regarding its cultural achievements. In this context, the importance of documenting such structures to create a substantial database cannot be emphasised enough. The Clock Tower of Dehradun, India is one such structure. There is a lack of sufficient information in the digital domain, which justified the need to carry out this study. Thus, an attempt has been made to gauge the possibilities of using open source 3D tools such as VSfM to quickly and easily obtain point clouds of an object and assess its quality. The photographs were collected using consumer grade cameras with reasonable effort to ensure overlap. The sparse reconstruction and dense reconstruction were carried out to generate a 3D point cloud model of the tower. A terrestrial laser scanner (TLS) was also used to obtain a point cloud of the tower. The point clouds obtained from the two methods were analyzed to understand the quality of the information present; TLS acquired point cloud being a benchmark to assess the VSfM point cloud. They were compared to analyze the point density and subjected to a plane-fitting test for sample flat portions on the structure. The plane-fitting test revealed the planarity of the point clouds. A Gauss distribution fit yielded a standard deviation of 0.002 and 0.01 for TLS and VSfM, respectively. For more insight, comparisons with Agisoft Photoscan results were also made.
- 5 July 2017
- Published by Copernicus GmbH in The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
- Vol. XLII-4/W2, 67-71
Open AccessCOMPARATIVE ANALYSIS OF 3D POINT CLOUDS GENERATED FROM A FREEWARE AND TERRESTRIAL LASER SCANNER
Keywords: Laser Scanning, Close Range Photogrammetry, Point clouds, Heritage, VSfM In the recent past, several heritage structures have faced destruction due to both human-made incidents and natural calamities that have caused a great loss to the human race regarding its cultural achievements. In this context, the importance of documenting such structures to create a substantial database cannot be emphasised enough. The Clock Tower of Dehradun, India is one such structure. There is a lack of sufficient information in the digital domain, which justified the need to carry out this study. Thus, an attempt has been made to gauge the possibilities of using open source 3D tools such as VSfM to quickly and easily obtain point clouds of an object and assess its quality. The photographs were collected using consumer grade cameras with reasonable effort to ensure overlap. The sparse reconstruction and dense reconstruction were carried out to generate a 3D point cloud model of the tower. A terrestrial laser scanner (TLS) was also used to obtain a point cloud of the tower. The point clouds obtained from the two methods were analyzed to understand the quality of the information present; TLS acquired point cloud being a benchmark to assess the VSfM point cloud. They were compared to analyze the point density and subjected to a plane-fitting test for sample flat portions on the structure. The plane-fitting test revealed the planarity of the point clouds. A Gauss distribution fit yielded a standard deviation of 0.002 and 0.01 for TLS and VSfM, respectively. For more insight, comparisons with Agisoft Photoscan results were also made.