Abhishek Rai
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Abhishek Rai has a total of 17 co-authors. Here are the top co-authors by number of shared publications:
- Aboelmagd Noureldin
Royal Military College of Canada
1 shared publication - Aditya Kumar Thakur
Department of Civil Engineering, Indian Institute of Technology Roorkee, Haridwar, India
1 shared publication - Aman Srivastava
Department of Civil Engineering, Indian Institute of Technology Roorkee, Haridwar, India
1 shared publication Ankit Sagar
Center for Nature-based Climate Solutions, Department of Geography, National University of Singapore, 117546, Singapore
1 shared publicationEslam Mounier
Queen’s University, Kingston, ON, Canada
1 shared publicationHristina Hristova
Sustainable Forestry, Forest Resources and Forest Management, Swiss Federal Institute for Forest, Snow and Landscape Research (WSL), Zurich, 8903, Switzerland
1 shared publication
Latest Publications
- Open Access
- 31 July 2025
- Published by Copernicus GmbH in The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
- Vol. XLVIII-G-2, 1255-1262
Investigation and Implementation of Multi-Stereo Camera System Integration for Robust Localization in Urban Environments
Urban environments are dynamic and complex, posing constant challenges for the localization and navigation of autonomous vehicles (AV). This demands more innovative sensor systems for effective autonomous navigation. Autonomous vehicles use sensors like LiDAR, cameras, and radar to traverse complicated urban environments with precision. These technologies have advantages in improving perception and localization, but they have their own shortcomings – LiDAR can be costly and falters under adverse weather conditions, cameras are sensitive to lighting conditions, and radars lack high-resolution details. Beyond these complexities, environmental conditions like signal-blocking skyscrapers, unpredictable obstacles, and the high costs of precision sensing add further convolution. A multi-sensor integrated solution can be a reliable option to overcome these challenges. Our work explores the use of a multi-stereo camera array that provides a 360° perception for localization in dense urban environments. We use computer vision algorithms to derive 3D point clouds from stereo-images and localize the cameras using a prior 3D map to balance cost and performance. We tested the system in Calgary’s urban setting with various lighting conditions and GNSS-denied zones. Our approach provided accurate localization in 85% of the cases we tested. The results demonstrate that our multi-stereo camera system can help to achieve robust localization in challenging urban situations. This approach offers a cost-effective alternative to LiDAR-based systems while ensuring adequate accuracy.
- 31 July 2025
- Published by Copernicus GmbH in The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
- Vol. XLVIII-G-2, 1255-1262
Open AccessInvestigation and Implementation of Multi-Stereo Camera System Integration for Robust Localization in Urban Environments
Urban environments are dynamic and complex, posing constant challenges for the localization and navigation of autonomous vehicles (AV). This demands more innovative sensor systems for effective autonomous navigation. Autonomous vehicles use sensors like LiDAR, cameras, and radar to traverse complicated urban environments with precision. These technologies have advantages in improving perception and localization, but they have their own shortcomings – LiDAR can be costly and falters under adverse weather conditions, cameras are sensitive to lighting conditions, and radars lack high-resolution details. Beyond these complexities, environmental conditions like signal-blocking skyscrapers, unpredictable obstacles, and the high costs of precision sensing add further convolution. A multi-sensor integrated solution can be a reliable option to overcome these challenges. Our work explores the use of a multi-stereo camera array that provides a 360° perception for localization in dense urban environments. We use computer vision algorithms to derive 3D point clouds from stereo-images and localize the cameras using a prior 3D map to balance cost and performance. We tested the system in Calgary’s urban setting with various lighting conditions and GNSS-denied zones. Our approach provided accurate localization in 85% of the cases we tested. The results demonstrate that our multi-stereo camera system can help to achieve robust localization in challenging urban situations. This approach offers a cost-effective alternative to LiDAR-based systems while ensuring adequate accuracy.
- Open Access
- 29 July 2025
- Published by Copernicus GmbH in The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
- Vol. XLVIII-G-2, 831-838
Development of the augmented reality app for forestry application: ForestAR
Augmented Reality (AR) is being widely applied across various fields, seamlessly integrating with real-world scenarios. In forestry, there is significant potential for the use of AR to improve our understanding of the datasets along with the physical environment of the surroundings. Remote Sensing is a powerful technology that generates highly accurate datasets, with Light Detection and Ranging (LiDAR) Remote Sensing technology being particularly effective in creating geometrically precise point cloud datasets. By integrating LiDAR datasets with AR applications, we can unlock numerous benefits. For one, the development of AR technologies in forestry can significantly improve the visualization, monitoring, management, and education of forest ecosystems by providing immersive digital experiences. This paper aims to demonstrate the potential of an AR application for visualizing different configurations of Terrestrial Laser Scanners (TLS) and Classified Point Cloud (CPC) of a forest plot. Such an AR approach offers a unique way to comprehensively examine forest structures. It also allows for the effective recreation, visualization, and presentation of an interactive environment of forest areas.
- 29 July 2025
- Published by Copernicus GmbH in The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
- Vol. XLVIII-G-2, 831-838
Open AccessDevelopment of the augmented reality app for forestry application: ForestAR
Augmented Reality (AR) is being widely applied across various fields, seamlessly integrating with real-world scenarios. In forestry, there is significant potential for the use of AR to improve our understanding of the datasets along with the physical environment of the surroundings. Remote Sensing is a powerful technology that generates highly accurate datasets, with Light Detection and Ranging (LiDAR) Remote Sensing technology being particularly effective in creating geometrically precise point cloud datasets. By integrating LiDAR datasets with AR applications, we can unlock numerous benefits. For one, the development of AR technologies in forestry can significantly improve the visualization, monitoring, management, and education of forest ecosystems by providing immersive digital experiences. This paper aims to demonstrate the potential of an AR application for visualizing different configurations of Terrestrial Laser Scanners (TLS) and Classified Point Cloud (CPC) of a forest plot. Such an AR approach offers a unique way to comprehensively examine forest structures. It also allows for the effective recreation, visualization, and presentation of an interactive environment of forest areas.
- Open Access
- 14 July 2025
- Published by Copernicus GmbH in ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
- Vol. X-G-2025, 855-862
Temporal and Spatial Variations of Nitrogen Dioxide Concentrations in Kolkata, India
Air pollution is a primary environmental concern in urban areas. This study examines the temporal and spatial variations in nitrogen dioxide (NO2) concentrations in Kolkata from 2019 to 2023 using Sentinel-5P satellite data. The application of statistical techniques, including Global Moran's I and Fast Fourier Transform (FFT), highlights changes in NO2 spatial distribution and identifies dominant periodicities. Yearly analysis reveals notable fluctuations in NO2 levels, with a significant decline of approximately 9.1% between 2019 and 2020, attributed to reduced vehicular and industrial activities during the COVID-19 lockdown. However, by 2023, NO2 concentrations had returned to pre-pandemic levels attributed to the resumption of economic activities. Spatial analysis reveals higher NO2 concentrations in central built-up areas, including Ballygunge, Bhowanipore, and Park Street. At the same time, peripheral regions such as Metiabruz and Behala show lower levels, likely due to vegetated areas. Global Correspondence values indicate significant shifts in NO2 distribution patterns over the study period. The pattern shifted during the COVID-19 pandemic but stabilized by 2023, aligning with pre-pandemic emission levels, but the monthly pattern was preserved. The FFT analysis reveals a dominant annual cycle with a frequency of 0.0833 cycles per month (12-month period) and an amplitude of 0.3520, along with a significant overall average component. Seasonal variations show higher concentrations in winter due to increased emissions and reduced levels in summer due to photolysis and monsoon rains. These findings underscore the importance of effective pollution management and continuous air quality monitoring to improve air quality in Kolkata.
- 14 July 2025
- Published by Copernicus GmbH in ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
- Vol. X-G-2025, 855-862
Open AccessTemporal and Spatial Variations of Nitrogen Dioxide Concentrations in Kolkata, India
Air pollution is a primary environmental concern in urban areas. This study examines the temporal and spatial variations in nitrogen dioxide (NO2) concentrations in Kolkata from 2019 to 2023 using Sentinel-5P satellite data. The application of statistical techniques, including Global Moran's I and Fast Fourier Transform (FFT), highlights changes in NO2 spatial distribution and identifies dominant periodicities. Yearly analysis reveals notable fluctuations in NO2 levels, with a significant decline of approximately 9.1% between 2019 and 2020, attributed to reduced vehicular and industrial activities during the COVID-19 lockdown. However, by 2023, NO2 concentrations had returned to pre-pandemic levels attributed to the resumption of economic activities. Spatial analysis reveals higher NO2 concentrations in central built-up areas, including Ballygunge, Bhowanipore, and Park Street. At the same time, peripheral regions such as Metiabruz and Behala show lower levels, likely due to vegetated areas. Global Correspondence values indicate significant shifts in NO2 distribution patterns over the study period. The pattern shifted during the COVID-19 pandemic but stabilized by 2023, aligning with pre-pandemic emission levels, but the monthly pattern was preserved. The FFT analysis reveals a dominant annual cycle with a frequency of 0.0833 cycles per month (12-month period) and an amplitude of 0.3520, along with a significant overall average component. Seasonal variations show higher concentrations in winter due to increased emissions and reduced levels in summer due to photolysis and monsoon rains. These findings underscore the importance of effective pollution management and continuous air quality monitoring to improve air quality in Kolkata.
- Open Access
- 14 December 2023
- Published by Copernicus GmbH in The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
- Vol. XLVIII-1/W, 1659-1666
SEMANTIC ENRICHMENT OF 3D POINT CLOUDS USING 2D IMAGE SEGMENTATION
3D point cloud segmentation is computationally intensive due to the lack of inherent structural information and the unstructured nature of the point cloud data, which hinders the identification and connection of neighboring points. Understanding the structure of the point cloud data plays a crucial role in obtaining a meaningful and accurate representation of the underlying 3D environment. In this paper, we propose an algorithm that builds on existing state-of-the-art techniques of 2D image segmentation and point cloud registration to enrich point clouds with semantic information. DeepLab2 with ResNet50 as backbone architecture trained on the COCO dataset is used for indoor scene semantic segmentation into several classes like wall, floor, ceiling, doors, and windows. Semantic information from 2D images is propagated along with other input data, i.e., RGB images, depth images, and sensor information to generate 3D point clouds with semantic information. Iterative Closest Point (ICP) algorithm is used for the pair-wise registration of consecutive point clouds and finally, optimization is applied using the pose graph optimization on the whole set of point clouds to generate the combined point cloud of the whole scene. 3D point cloud of the whole scene contains pseudo-color information which denotes the semantic class to which each point belongs. The proposed methodology use an off-the-shelf 2D semantic segmentation deep learning model to semantically segment 3D point clouds collected using handheld mobile LiDAR sensor. We demonstrate a comparison of the accuracy achieved compared to a manually segmented point cloud on an in-house dataset as well as a 2D3DS benchmark dataset.
- 14 December 2023
- Published by Copernicus GmbH in The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
- Vol. XLVIII-1/W, 1659-1666
Open AccessSEMANTIC ENRICHMENT OF 3D POINT CLOUDS USING 2D IMAGE SEGMENTATION
3D point cloud segmentation is computationally intensive due to the lack of inherent structural information and the unstructured nature of the point cloud data, which hinders the identification and connection of neighboring points. Understanding the structure of the point cloud data plays a crucial role in obtaining a meaningful and accurate representation of the underlying 3D environment. In this paper, we propose an algorithm that builds on existing state-of-the-art techniques of 2D image segmentation and point cloud registration to enrich point clouds with semantic information. DeepLab2 with ResNet50 as backbone architecture trained on the COCO dataset is used for indoor scene semantic segmentation into several classes like wall, floor, ceiling, doors, and windows. Semantic information from 2D images is propagated along with other input data, i.e., RGB images, depth images, and sensor information to generate 3D point clouds with semantic information. Iterative Closest Point (ICP) algorithm is used for the pair-wise registration of consecutive point clouds and finally, optimization is applied using the pose graph optimization on the whole set of point clouds to generate the combined point cloud of the whole scene. 3D point cloud of the whole scene contains pseudo-color information which denotes the semantic class to which each point belongs. The proposed methodology use an off-the-shelf 2D semantic segmentation deep learning model to semantically segment 3D point clouds collected using handheld mobile LiDAR sensor. We demonstrate a comparison of the accuracy achieved compared to a manually segmented point cloud on an in-house dataset as well as a 2D3DS benchmark dataset.