Real-time GeoAI for high-resolution mapping and segmentation of arctic permafrost features: the case of ice-wedge polygons
Open accessThe DOI matches its Crossref record; 18 of the 18 references it cites were found in a registry or an index.
Abstract
This paper introduces a real-time GeoAI workflow for large-scale image analysis and the segmentation of Arctic permafrost features at a fine-granularity. Very high-resolution (0.5m) commercial imagery is used in this analysis. To achieve real-time prediction, our workflow employs a lightweight, deep learning-based instance segmentation model, SparseInst, which introduces and uses Instance Activation Maps to accurately locate the position of objects within the image scene. Experimental results show that the model can achieve better accuracy of prediction at a much faster inference speed than the popular Mask-RCNN model.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“Through data-driven analysis, deep learning models have demonstrated the capability to recognize distinctive flood patterns and subtle flood boundaries, even in regions under complex and rapidly changing conditions.”
From Assessment of a new GeoAI foundation model for flood inundation mapping (2023) · cites this paper as [24]
Citations
Cites
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474 · cited 2×
“Among various deep neural network (DNN) architectures, ResNet-50 enjoys a good trade-off between accuracy and model complexity so to support real-time applications [3].”
MethodFeature Pyramid Networks for Object Detection
Lin, T.-Y., Dollar, P., Girshick, R., He, K., Hariharan, B., & Belongie, S. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.106 · Cited by 30,939 · cited 1×
“The fusion first follows the feature pyramid network (FPN) [14] to use a top-down pathway for building semantic-rich features.”
MethodHe, K., Gkioxari, G., Dollar, P., & Girshick, R. (2017) · 2017 IEEE International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2017.322 · Cited by 29,137
Zhao, H., Shi, J., Qi, X., Wang, X., & Jia, J. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.660 · Cited by 16,438 · cited 1×
“To further enhance the scale information, the last feature map (C3) also undergoes a pyramid pooling operation [23] to increase the global contextual information without increasing the size of the feature maps.”
MethodClimate change and the permafrost carbon feedback
Schuur, E. A. G., McGuire, A. D., Schädel, C., Grosse, G., Harden, J. W., Hayes, D. J., Hugelius, G., Koven, C. D., Kuhry, P., Lawrence, D. M., et al. (2015) · Nature · doi:10.1038/nature14338 · Cited by 3,746 · cited 1×
“At a global scale, the thawing of Arctic permafrost will result in the release of an immense amount of carbon dioxide and methane, exaggerating the greenhouse effect and global warming through complex feedback mechanisms [17].”
BackgroundYOLACT: Real-Time Instance Segmentation
Bolya, D., Zhou, C., Xiao, F., & Lee, Y. J. (2019) · 2019 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2019.00925 · Cited by 2,536 · cited 1×
“Recent real-time instance segmentation models [4, 16] fuse multi-scale information into a single feature map to reduce both prediction and post-processing time.”
BackgroundScale-Aware Trident Networks for Object Detection
Li, Y., Chen, Y., Wang, N., & Zhang, Z.-X. (2019) · 2019 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2019.00615 · Cited by 1,087 · cited 1×
“Conventional approaches use multi-scale features with multi-level predictions [13] for segmenting objects at different scales [20].”
BackgroundBenchmark analysis of representative deep neural network architectures
Bianco, S., Cadene, R., Celona, L., & Napoletano, P. (2018) · IEEE Access · doi:10.1109/access.2018.2877890 · Cited by 838 · cited 1×
“Among various deep neural network (DNN) architectures, ResNet-50 enjoys a good trade-off between accuracy and model complexity so to support real-time applications [3].”
BackgroundSparse Instance Activation for Real-Time Instance Segmentation
Cheng, T., Wang, X., Chen, S., Zhang, W., Zhang, Q., Huang, C., Zhang, Z., & Liu, W. (2022) · 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr52688.2022.00439 · Cited by 210 · cited 1×
“This paper aims to achieve this goal by integrating a novel real-time instance segmentation model, SparseInst [5], in our automated permafrost feature mapping pipeline.”
MethodChanging permafrost in a warming world and feedbacks to the Earth system
Grosse, G., Goetz, S., McGuire, A. D., Romanovsky, V. E., & Schuur, E. A. G. (2016) · Environmental Research Letters · doi:10.1088/1748-9326/11/4/040201 · Cited by 193 · cited 1×
“Permafrost degradation also increases rates of coastal erosion, wildfires, and flooding, which may further accelerate the thawing process and make the Arctic ecosystem even more vulnerable to climate change [6].”
Background
Research works printed but not linked: 3
- Jan Hjort , Olli Karjalainen , Juha Aalto , Sebastian Westermann , Vladimir E Romanovsky , Frederick E Nelson , Bernd Etzelmüller , and Miska Luoto . 2018. Degrading permafrost puts Arctic infrastructure at risk by mid-century. Nature communications 9, 1 ( 2018 ), 1--9. Jan Hjort, Olli Karjalainen, Juha Aalto, Sebastian Westermann, Vladimir E Romanovsky, Frederick E Nelson, Bernd Etzelmüller, and Miska Luoto. 2018. Degrading permafrost puts Arctic infrastructure at risk by mid-century. Nature communications 9, 1 (2018), 1--9.
- Hughes Perreault , Guillaume-Alexandre Bilodeau , Nicolas Saunier , and Maguelonne Héritier . 2021. CenterPoly: Real-Time Instance Segmentation Using Bounding Polygons. arXiv:2108.08923 [cs] (Sept . 2021 ). arXiv:2108.08923 [cs] Hughes Perreault, Guillaume-Alexandre Bilodeau, Nicolas Saunier, and Maguelonne Héritier. 2021. CenterPoly: Real-Time Instance Segmentation Using Bounding Polygons. arXiv:2108.08923 [cs] (Sept. 2021). arXiv:2108.08923 [cs] · arXiv
- Xinlong Wang , Rufeng Zhang , Tao Kong , Lei Li , and Chunhua Shen . 2020 . SOLOv2: Dynamic and Fast Instance Segmentation. In Advances in Neural Information Processing Systems, H. Larochelle, M. Ranzato, R. Hadsell, M. F. Balcan, and H . Lin (Eds.) , Vol. 33 . Curran Associates, Inc., 17721--17732. Xinlong Wang, Rufeng Zhang, Tao Kong, Lei Li, and Chunhua Shen. 2020. SOLOv2: Dynamic and Fast Instance Segmentation. In Advances in Neural Information Processing Systems, H. Larochelle, M. Ranzato, R. Hadsell, M. F. Balcan, and H. Lin (Eds.), Vol. 33. Curran Associates, Inc., 17721--17732.
Cited by
Hsu, C.-Y., & Li, W. (2023) · International Journal of Geographical Information Science · doi:10.1080/13658816.2023.2191256 · Cited by 75
“…2018, Helber et al. 2019, Li and Hsu 2020, Wang and Li 2021, Hsu et al. 2021), weather forecasting and nowcasting (Zhang et al. 2019), extreme climate event detection (Kurth et al. 2018), and neighborhood property quantification and change monitoring (Gebru et al. 2017, Koo et al. 2022, Li 2022a).”
BackgroundMethodSelf-citationLi, W., Hsu, C.-Y., Wang, S., Yang, Y., Lee, H., Liljedahl, A., Witharana, C., Yang, Y., Rogers, B. M., Arundel, S. T., et al. (2024) · Remote Sensing · doi:10.3390/rs16050797 · Cited by 52
Self-citationHsu, C.-Y., Li, W., & Wang, S. (2024) · International Journal of Geographical Information Systems · doi:10.1080/13658816.2024.2397441 · Cited by 49
Self-citationAssessment of a new GeoAI foundation model for flood inundation mapping
Li, W., Lee, H., Wang, S., Hsu, C.-Y., & Arundel, S. T. (2023) · arXiv · doi:10.48550/arxiv.2309.14500 · Cited by 40
Self-citationAssessment of a new GeoAI foundation model for flood inundation mapping
Li, W., Lee, H., Wang, S., Hsu, C.-Y., & Arundel, S. T. (2023) · Proceedings of the 6th ACM SIGSPATIAL International Workshop on AI for Geographic Knowledge Discovery · doi:10.1145/3615886.3627747 · Cited by 39
Self-citationA Multiscale Vision Transformer-Based Multimodal GeoAI Model for Mapping Arctic Permafrost Thaw
Li, W., Hsu, C.-Y., Wang, S., Gu, Z., Yang, Y., Rogers, B. M., & Liljedahl, A. (2025) · IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing · doi:10.1109/jstars.2025.3564310 · Cited by 14
“…in Geospatial Artificial Intelligence (GeoAI) [11] and Deep Learning (DL) have led to new models demonstrating outstanding performance in various computer vision tasks, including object detection [12], [13], [14], [15], [16], [17], and semantic and instance segmentation [12], [18], [19].”
Self-citationMueller, M. M., Dietenberger, S., Nestler, M., Dubois, C., Kaiser, S., Lenz, J., Langer, M., Fritz, O., Marx, S., & Thiel, C. (2024) · IEEE International Geoscience and Remote Sensing Symposium (IGARSS) · doi:10.1109/igarss53475.2024.10642540 · Cited by 3
“Traditionally, remote sensing technologies like satellite or airplane-based imaging have been employed to observe these changes [1 – 3].”
MethodEvaluation of the Potential of Very-High-Resolution Satellite Imagery in Large-Scale Mapping
Afa, I., Labbaci, A., Ghazouani, L. E., & Radoine, H. (2026) · Remote Sensing · doi:10.3390/rs18091421 · Cited by 2
“New datasets, such as GLH-Water, enable precise extraction of surface water features, while deep learning models trained on aerial LiDAR have achieved strong results in canopy height estimation from RGB satel-https lite images [10,11].”
Evaluation of Explainable Ai Models in Analysis of Transferable Features of Geothermal IndicatorsPreprint
Demir, E., Düzgün, Ş., & Çavur, M. (2024) · SSRN Electronic Journal · doi:10.2139/ssrn.4897692
Trongtirakul, T., Agaian, S. С., Chaudhuri, S. S., Djemal, K., & Feiz, A. A. (2026) · Information · doi:10.3390/info17040373
Getting Started in Big Data Remote Sensing
Bunting, E., Wanyama, D., & Tolmanov, V. (2026) · Elsevier eBooks · doi:10.1016/b978-0-443-26741-3.00006-4
Analisis Banjir Di Sungai Ambawang Menggunakan Topography Wetness Index
Purwanto, A., & Paiman, P. (2025) · JURNAL SAINS INFORMASI GEOGRAFIS · doi:10.31314/jsig.v8i2.4602 · Cited by 1
Permafrost thaw follows spatially organized, nonlinear trajectories in ice-wedge terrainPreprint
Braun, K. N., May, J. A., & Andresen, C. G. (2026) · doi:10.5194/egusphere-2026-5192
Metadata
- Authors
- Li, W., Hsu, C.-Y., Wang, S., Witharana, C., & Liljedahl, A.
- Affiliations
- Arizona State University (US)
- University of Connecticut (US)
- Woodwell Climate Research Center (US)
- Venue
- Proceedings of the 5th ACM SIGSPATIAL International Workshop on AI for Geographic Knowledge Discovery
- Publisher
- ACM
- Published
- 2022-11
- Pages
- 62-65
- OpenAlex
- W4308986509
- Cited by
- 9
- References
- 18
- Citations by year
- 2026: 1 · 2025: 1 · 2024: 5 · 2023: 2
- FWCI
- 0.73
- Type
- conference paper
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- ACM SIGSPATIAL International Workshop on AI for Geographic Knowledge Discovery (GeoAI) (publishedVersion)
- arXiv (Cornell University) (submittedVersion, cc-by)
- arXiv (Cornell University) (cc-by)
- Funders
- National Science Foundation (2120943)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Remote-Sensing Image Classification
- Physical Sciences › Earth and Planetary Sciences › Atmospheric Science › Climate change and permafrost
- Keywords
- Arctic permafrost, instance segmentation, real-time inference, very high resolution imagery, lightweight deep learning, Mask R-CNN comparison
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.