Jaya Sreevalsan-Nair
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Jaya Sreevalsan-Nair has a total of 98 co-authors. Here are the top co-authors by number of shared publications:
- Alberto Moreira
German Aerospace Center (DLR), Microwaves and Radar Institute, Weling, Germany
4 shared publications - Francesca Bovolo
Fondazione Bruno Kessler, Trento, Italy
4 shared publications Reddy Rani Vangimalla
BITS Pilani, Rajasthan, India
4 shared publications- Reddy Rani Vangimalla
Graphics-Visualization-Computing Laboratory (GVCL) and the EHRC, International Institute of Information Technology, Bengaluru, Karnataka, India
4 shared publications - Urvakhsh Meherwan Mehta
Department of Psychiatry, National Institute of Mental Health and Neurosciences
4 shared publications - Antonio Plaza
University of Extremadura
3 shared publications
Latest Publications
- 1 October 2026
- Published by Springer Nature
- p. 588-602
SeeIntent: A Real-Time Multimodal Intent Recognition System in XR Using Spatio-Temporal LLM
- 1 October 2026
- Published by Springer Nature
- p. 588-602
SeeIntent: A Real-Time Multimodal Intent Recognition System in XR Using Spatio-Temporal LLM
- Open Access
- 14 April 2026
- Published by Springer Nature in Archives of Computational Methods in Engineering
- Vol. 33 (6) , 1-36
Road and Building Reconstruction from 3D LiDAR Point Clouds: A Scoping Review
Recent advancements in semantic 3D city building reconstruction have achieved high levels of geometric detail, enabling a wide range of applications in urban planning, digital twins, and smart city systems. However, the state of the art (SOTA) has three glaring gaps. Firstly, building and road reconstruction are often treated as decoupled tasks, overlooking the mutual benefits of integrating both within a unified modeling framework. The integration is beneficial, as roads could serve as reliable spatial priors that aid in the determination of the optimal layout of buildings. Secondly, despite the availability of large-scale point clouds from airborne and mobile LiDAR (Light Detection and Ranging) systems, most existing workflows remain manual or semi-automated, relying on rule-based modeling and expert intervention. Finally, there is a lack of open datasets for an integrated 3D building-road reconstruction. A scoping review of the existing work is much needed to identify these gaps. This comprehensive unifying review of existing techniques includes both building and road reconstruction from point clouds, spanning both parametric and data-driven approaches. Further, a few key building and road reconstruction methods are implemented on an example open dataset to identify challenges. This pipeline treats road modeling as a prior; thus, it is implemented before the reconstruction of buildings. This example highlights key challenges, such as pose correction, ambiguity in footprint estimation from MLS data, and semantic inconsistencies in the reconstructed meshes, that need to be studied further.
- 14 April 2026
- Published by Springer Nature in Archives of Computational Methods in Engineering
- Vol. 33 (6) , 1-36
Open AccessRoad and Building Reconstruction from 3D LiDAR Point Clouds: A Scoping Review
Recent advancements in semantic 3D city building reconstruction have achieved high levels of geometric detail, enabling a wide range of applications in urban planning, digital twins, and smart city systems. However, the state of the art (SOTA) has three glaring gaps. Firstly, building and road reconstruction are often treated as decoupled tasks, overlooking the mutual benefits of integrating both within a unified modeling framework. The integration is beneficial, as roads could serve as reliable spatial priors that aid in the determination of the optimal layout of buildings. Secondly, despite the availability of large-scale point clouds from airborne and mobile LiDAR (Light Detection and Ranging) systems, most existing workflows remain manual or semi-automated, relying on rule-based modeling and expert intervention. Finally, there is a lack of open datasets for an integrated 3D building-road reconstruction. A scoping review of the existing work is much needed to identify these gaps. This comprehensive unifying review of existing techniques includes both building and road reconstruction from point clouds, spanning both parametric and data-driven approaches. Further, a few key building and road reconstruction methods are implemented on an example open dataset to identify challenges. This pipeline treats road modeling as a prior; thus, it is implemented before the reconstruction of buildings. This example highlights key challenges, such as pose correction, ambiguity in footprint estimation from MLS data, and semantic inconsistencies in the reconstructed meshes, that need to be studied further.
- Open Access
- 9 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 22882-22905
Adversarial Robustness of Deep Learning Models for Inland Water Body Segmentation From SAR Images
Segmentation of inland water bodies from Synthetic Aperture Radar (SAR) images is crucial for several applications, such as flood mapping and monitoring. While SAR sensors provide high-resolution, all-weather imagery, distinguishing water from water-like surfaces remains challenging due to the complex geometry of inland regions. Semantic segmentation models, including deep learning and transformer-based approaches, are increasingly used for land-water segmentation. However, manual annotation—commonly used to generate water masks as ground truth—is prone to label noise, particularly near land-water boundaries. Currently, there are no datasets and studies that capture such human annotation errors to help train models to be robust to such noise. In this work, we simulate manual annotation errors as adversarial attacks using morphological operations and study the robustness of various segmentation models on our curated dataset. We compare the robustness of the chosen models to these morphological poisoning attacks, and also create a publicly available dataset with these adversarial examples. Our results show that U-Net can tolerate a certain level of annotation corruption before a sharp performance drop, while other models exhibit a gradual decline, with Mask2Former being an anomaly due to the smaller dataset size and its mask-classification paradigm. We validate the statistical significance and robustness of the results through paired t-tests and 95% confidence intervals. We also discuss a few anomalies in the dataset itself and how the U-Net model is immune to these anomalies. These findings highlight the critical role of annotation quality in determining model effectiveness. We also release our code and a new dataset containing these adversarial examples for robust model training. (GitHub: https://github.com/GVCL/IWSeg-SAR-Poison.git)
- 9 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 22882-22905
Open AccessAdversarial Robustness of Deep Learning Models for Inland Water Body Segmentation From SAR Images
Segmentation of inland water bodies from Synthetic Aperture Radar (SAR) images is crucial for several applications, such as flood mapping and monitoring. While SAR sensors provide high-resolution, all-weather imagery, distinguishing water from water-like surfaces remains challenging due to the complex geometry of inland regions. Semantic segmentation models, including deep learning and transformer-based approaches, are increasingly used for land-water segmentation. However, manual annotation—commonly used to generate water masks as ground truth—is prone to label noise, particularly near land-water boundaries. Currently, there are no datasets and studies that capture such human annotation errors to help train models to be robust to such noise. In this work, we simulate manual annotation errors as adversarial attacks using morphological operations and study the robustness of various segmentation models on our curated dataset. We compare the robustness of the chosen models to these morphological poisoning attacks, and also create a publicly available dataset with these adversarial examples. Our results show that U-Net can tolerate a certain level of annotation corruption before a sharp performance drop, while other models exhibit a gradual decline, with Mask2Former being an anomaly due to the smaller dataset size and its mask-classification paradigm. We validate the statistical significance and robustness of the results through paired t-tests and 95% confidence intervals. We also discuss a few anomalies in the dataset itself and how the U-Net model is immune to these anomalies. These findings highlight the critical role of annotation quality in determining model effectiveness. We also release our code and a new dataset containing these adversarial examples for robust model training. (GitHub: https://github.com/GVCL/IWSeg-SAR-Poison.git)
- Open Access
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 169899-169931
Consensus Clustering of Free-Viewing Gaze Data: New Insights Into Human-Information Interaction
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 14, 169899-169931
Open AccessConsensus Clustering of Free-Viewing Gaze Data: New Insights Into Human-Information Interaction
- Open Access
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 19, 35383-35416
FL-MHSM: Spatially Adaptive Fusion and Ensemble Learning for Flood–Landslide Multihazard Susceptibility Mapping at Regional Scale
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 19, 35383-35416
Open AccessFL-MHSM: Spatially Adaptive Fusion and Ensemble Learning for Flood–Landslide Multihazard Susceptibility Mapping at Regional Scale
- 17 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Magazine
- Vol. 13 (4) , 444-454
IEEE Geoscience and Remote Sensing Society Awards Presented at the IEEE International Geoscience and Remote Sensing Symposium 2025 Awards Reception and Ceremony [Conference Reports]
Provides society information that may include news, reviews or technical notes that should be of interest to practitioners and researchers.
- 17 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Magazine
- Vol. 13 (4) , 444-454
IEEE Geoscience and Remote Sensing Society Awards Presented at the IEEE International Geoscience and Remote Sensing Symposium 2025 Awards Reception and Ceremony [Conference Reports]
Provides society information that may include news, reviews or technical notes that should be of interest to practitioners and researchers.
- 17 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Magazine
- Vol. 13 (4) , 430-443
IGARSS 2025 in Brisbane, Australia: Impressions of the First Days [Conference Reports]
Provides society information that may include news, reviews or technical notes that should be of interest to practitioners and researchers.
- 17 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Geoscience and Remote Sensing Magazine
- Vol. 13 (4) , 430-443
IGARSS 2025 in Brisbane, Australia: Impressions of the First Days [Conference Reports]
Provides society information that may include news, reviews or technical notes that should be of interest to practitioners and researchers.
- 15 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Biology and Bioinformatics
- Vol. 22 (6) , 3550-3556
Network-Based Diseasome Construction From Multi-Omics Data and RadTrix Visualization
Gene-gene association studies are essential for genomic analysis in network medicine. Multi-omics datasets of multiple disease phenotypes are used to construct a disease-gene network or diseasome, which gives a set of gene nodes much larger than the set of disease phenotypes. The resulting network is an unbalanced bipartite graph, which leads to cluttered visualizations. In the state of the art, a relatively small set of top-ranked genes is used for exploratory analysis to resolve the issue of unbalancedness, but that approach usually limits the scope of the analysis. Hence, we use a novel hybrid graph layout, RadTrix, for representing an unbalanced diseasome. We design a novel browser-based tool, RadTrixVis, to integrate the diseasome construction from multi-omics data and its RadTrix-based visualization. We further propose novel gene ranking and filtering techniques to improve the efficiency of our tool. We demonstrate the use of RadTrixVis through a case study of seven oncological phenotypes from The Cancer Genome Atlas (TCGA). Our results show the effectiveness of RadTrixVis in the exploratory study of the diseasome.
- 15 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Biology and Bioinformatics
- Vol. 22 (6) , 3550-3556
Network-Based Diseasome Construction From Multi-Omics Data and RadTrix Visualization
Gene-gene association studies are essential for genomic analysis in network medicine. Multi-omics datasets of multiple disease phenotypes are used to construct a disease-gene network or diseasome, which gives a set of gene nodes much larger than the set of disease phenotypes. The resulting network is an unbalanced bipartite graph, which leads to cluttered visualizations. In the state of the art, a relatively small set of top-ranked genes is used for exploratory analysis to resolve the issue of unbalancedness, but that approach usually limits the scope of the analysis. Hence, we use a novel hybrid graph layout, RadTrix, for representing an unbalanced diseasome. We design a novel browser-based tool, RadTrixVis, to integrate the diseasome construction from multi-omics data and its RadTrix-based visualization. We further propose novel gene ranking and filtering techniques to improve the efficiency of our tool. We demonstrate the use of RadTrixVis through a case study of seven oncological phenotypes from The Cancer Genome Atlas (TCGA). Our results show the effectiveness of RadTrixVis in the exploratory study of the diseasome.