pith. sign in

arxiv: 2501.05281 · v2 · submitted 2025-01-09 · 💻 cs.CV · cs.LG

Comparison Study: Glacier Calving Front Delineation in Synthetic Aperture Radar Images With Deep Learning

Nora Gourmelon , Konrad Heidler , Erik Loebel , Daniel Cheng , Julian Klink , Anda Dong , Fei Wu , Noah Maul
show 7 more authors
Moritz Koch Marcel Dreier Dakota Pyles Thorsten Seehaus Matthias Braun Andreas Maier Vincent Christlein
This is my paper

Pith reviewed 2026-05-23 05:51 UTC · model grok-4.3

classification 💻 cs.CV cs.LG
keywords glacier calving frontdeep learningSAR imageryimage delineationbenchmark studyremote sensingsea level riseglacier monitoring
0
0 comments X

The pith

Deep learning models for glacier calving front delineation in SAR images reach errors of 221 m while human annotators stay within 38 m.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper benchmarks several deep learning approaches for automatically tracing glacier calving fronts in synthetic aperture radar satellite images. The central finding is that the best models still produce maximum deviations of 221 meters from reference lines, whereas human experts deviate by only 38 meters. The work argues that this performance gap means current automated systems cannot yet replace manual mapping for reliable long-term monitoring. Such monitoring matters because accurate front positions feed directly into projections of glacier mass loss and sea level rise. The authors conclude that additional research is required to close the accuracy difference.

Core claim

Benchmark tests on SAR imagery show that deep learning systems for calving front delineation produce errors reaching 221 m, while human annotators produce deviations of only 38 m, indicating that present models fall short of the precision needed for operational use.

What carries the argument

Direct error comparison between deep learning outputs and human annotations on the same set of SAR images, measured in meters of deviation from reference front lines.

If this is right

  • Current deep learning outputs require human correction before they can support sea level rise calculations.
  • SAR-based monitoring of calving fronts will remain partly manual until model errors drop closer to human levels.
  • Operational glacier tracking systems will need accuracy improvements before scaling to many more glaciers worldwide.
  • The 38 m human benchmark sets a concrete performance target for future model development.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • If models reach human-level accuracy, fully automated processing could enable daily or weekly front updates across large regions instead of sporadic manual mapping.
  • The benchmark dataset and error protocol could be reused to test new architectures or training strategies on the same problem.
  • Similar accuracy gaps may exist in related remote-sensing tasks such as coastline or ice-shelf edge delineation, suggesting the comparison method has broader utility.
  • Repeating the study with newer model families could reveal whether the 221 m ceiling is architecture-specific or inherent to the data.

Load-bearing premise

The 38 m human deviation provides a trustworthy reference standard and the evaluated deep learning models represent the current best available performance for this task.

What would settle it

A new deep learning model tested on the identical SAR dataset that achieves a maximum error below 38 m would show the reported gap can be closed with existing techniques.

Figures

Figures reproduced from arXiv: 2501.05281 by Anda Dong, Andreas Maier, Dakota Pyles, Daniel Cheng, Erik Loebel, Fei Wu, Julian Klink, Konrad Heidler, Marcel Dreier, Matthias Braun, Moritz Koch, Noah Maul, Nora Gourmelon, Thorsten Seehaus, Vincent Christlein.

Figure 1
Figure 1. Figure 1: Overview of MDEs with confidence intervals alongside the number of images with no predicted front for all 22 DL systems and the comparisons [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Predicted calving fronts of the five best-performing DL systems for an image of the Mapple Glacier taken on 24th October 2008 by the TerraSAR-X [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Predicted calving fronts of the five best-performing DL systems for an image of the Columbia Glacier taken on 8th September 2017 by the Sentinel-1 [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Visualizations for all ten annotations by humans ( [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Predicted calving fronts of all 22 DL systems for an image of the Columbia Glacier taken on 2nd January 2012 by the TanDEM-X satellite. [PITH_FULL_IMAGE:figures/full_fig_p020_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Predicted calving fronts of all 22 DL systems for an image of the Mapple Glacier taken on 13th October 2008 by the TerraSAR-X satellite. [PITH_FULL_IMAGE:figures/full_fig_p021_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Overview of the annotators’ QGIS proficiency and their knowledge of glaciers. [PITH_FULL_IMAGE:figures/full_fig_p025_7.png] view at source ↗
read the original abstract

Continuous monitoring of glacier calving fronts is essential for sea level rise projections. This study benchmarks Deep Learning systems for front delineation in Synthetic Aperture Radar imagery. While Deep Learning systems exhibit errors up to 221 m, human annotators deviate by only 38 m, underscoring the need for further research.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The manuscript is a benchmark study comparing deep learning approaches for delineating glacier calving fronts in SAR imagery. It reports DL errors reaching 221 m against a human annotator deviation of 38 m and concludes that further research is required to improve automated monitoring for sea-level-rise applications.

Significance. If the tested models are representative of current best performance, the reported gap supplies a useful empirical reference point for the limitations of DL in this remote-sensing task and could motivate targeted improvements in segmentation accuracy.

major comments (2)
  1. [Abstract/Methods] Abstract and Methods: the manuscript supplies no information on dataset size, annotation protocol, precise definition of the error metric, or statistical significance testing; without these the headline figures of 221 m (DL) and 38 m (human) cannot be interpreted reliably.
  2. [Results/Methods] Results/Methods: the evaluated DL systems are not shown to constitute current state-of-the-art architectures or training regimes for SAR calving-front segmentation; if stronger published methods close the gap to the 38 m human reference, both the magnitude of the reported performance gap and the call for further research are weakened.
minor comments (1)
  1. The title does not indicate the number of DL methods compared or the geographic coverage of the SAR dataset.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive comments. We address each major comment below and have revised the manuscript accordingly.

read point-by-point responses

  1. Referee: [Abstract/Methods] Abstract and Methods: the manuscript supplies no information on dataset size, annotation protocol, precise definition of the error metric, or statistical significance testing; without these the headline figures of 221 m (DL) and 38 m (human) cannot be interpreted reliably.

    Authors: We agree that these details are essential for reliable interpretation. The revised manuscript now includes explicit information on dataset size, annotation protocol, the precise definition of the error metric, and statistical significance testing results in both the Abstract and Methods sections. revision: yes

  2. Referee: [Results/Methods] Results/Methods: the evaluated DL systems are not shown to constitute current state-of-the-art architectures or training regimes for SAR calving-front segmentation; if stronger published methods close the gap to the 38 m human reference, both the magnitude of the reported performance gap and the call for further research are weakened.

    Authors: The evaluated systems represent commonly used architectures for this task. We acknowledge they are not presented as the absolute latest SOTA. The revised manuscript adds a discussion of more recent methods and their potential impact on the gap, while maintaining that the reported performance difference still supports the need for further research. revision: yes

Circularity Check

0 steps flagged

Empirical benchmark study with no derivations or circular steps

full rationale

This is a comparison study reporting measured errors (DL up to 221 m, humans 38 m) on SAR imagery. No equations, first-principles derivations, fitted parameters renamed as predictions, or self-citation chains appear in the provided abstract or described content. The central claim rests on direct empirical comparison rather than any reduction to inputs by construction, satisfying the criteria for a self-contained non-circular result.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Empirical benchmarking study with no mathematical derivations, free parameters, axioms, or invented entities.

pith-pipeline@v0.9.0 · 5612 in / 933 out tokens · 31407 ms · 2026-05-23T05:51:36.591941+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

62 extracted references · 62 canonical work pages · 3 internal anchors

  1. [1]

    Mass balance of the greenland and antarctic ice sheets from 1992 to 2020,

    I. N. Otosaka, A. Shepherd, E. R. Ivins, N.-J. Schlegel, C. Amory, M. R. van den Broeke, M. Horwath, I. Joughin, M. D. King, G. Krinner, S. Nowicki, A. J. Payne, E. Rignot, T. Scambos, K. M. Simon, B. E. Smith, L. S. Sørensen, I. Velicogna, P. L. Whitehouse, G. A, C. Agosta, A. P. Ahlstrøm, A. Blazquez, W. Colgan, M. E. Engdahl, X. Fettweis, R. Forsberg, ...

    work page 1992

  2. [2]

    Accelerated global glacier mass loss in the early twenty-first century,

    R. Hugonnet, R. McNabb, E. Berthier, B. Menounos, C. Nuth, L. Girod, D. Farinotti, M. Huss, I. Dussaillant, F. Brun, and A. K ¨a¨ab, “Accelerated global glacier mass loss in the early twenty-first century,” Nature, vol. 592, no. 7856, pp. 726–731, 2021

    work page 2021

  3. [3]

    Greenland ice sheet mass balance: a review,

    S. A. Khan, A. Aschwanden, A. A. Bjørk, J. Wahr, K. K. Kjeldsen, and K. H. Kjær, “Greenland ice sheet mass balance: a review,” Reports on progress in physics. Physical Society (Great Britain) , vol. 78, no. 4, p. 046801, 2015

    work page 2015

  4. [4]

    Mass balance of the antarctic ice sheet from 1992 to 2017,

    A. Sheperd, E. Ivins, E. Rignot, B. Smith, M. van den Broeke, I. Velicogna, P. Whitehouse, K. Briggs, and I. Joughin, “Mass balance of the antarctic ice sheet from 1992 to 2017,” Nature, vol. 558, no. 7709, pp. 219–222, 2018

    work page 1992

  5. [5]

    Impact of frontal ablation on the ice thickness estimation of marine-terminating glaciers in Alaska,

    B. Recinos, F. Maussion, T. Rothenpieler, and B. Marzeion, “Impact of frontal ablation on the ice thickness estimation of marine-terminating glaciers in Alaska,” The Cryosphere , vol. 13, no. 10, pp. 2657–2672, Oct. 2019. [Online]. Available: https: //tc.copernicus.org/articles/13/2657/2019/

    work page 2019

  6. [6]

    Global glacier change in the 21st century: Every increase in temperature matters,

    D. R. Rounce, R. Hock, F. Maussion, R. Hugonnet, W. Kochtitzky, M. Huss, E. Berthier, D. Brinkerhoff, L. Compagno, L. Copland, D. Farinotti, B. Menounos, and R. W. McNabb, “Global glacier change in the 21st century: Every increase in temperature matters,” Science, vol. 379, no. 6627, pp. 78–83, 2023

    work page 2023

  7. [7]

    The mechanisms behind jakobshavn isbræ’s acceleration and mass loss: A 3-d thermomechanical model study,

    J. H. Bondzio, M. Morlighem, H. Seroussi, T. Kleiner, M. R ¨uckamp, J. Mouginot, T. Moon, E. Y . Larour, and A. Humbert, “The mechanisms behind jakobshavn isbræ’s acceleration and mass loss: A 3-d thermomechanical model study,” Geophysical Research Letters , vol. 44, no. 12, pp. 6252–6260, 2017. [Online]. Available: https: //agupubs.onlinelibrary.wiley.co...

    work page doi:10.1002/2017gl073309 2017

  8. [8]

    Understanding and modelling rapid dynamic changes of tidewater outlet glaciers: Issues and implications,

    A. Vieli and F. M. Nick, “Understanding and modelling rapid dynamic changes of tidewater outlet glaciers: Issues and implications,” Surveys in Geophysics, vol. 32, no. 4, pp. 437–458, 2011

    work page 2011

  9. [9]

    Automated extraction of antarctic glacier and ice shelf fronts from sentinel-1 imagery using deep learning,

    C. A. Baumhoer, A. J. Dietz, C. Kneisel, and C. Kuenzer, “Automated extraction of antarctic glacier and ice shelf fronts from sentinel-1 imagery using deep learning,” Remote Sensing, vol. 11, no. 21, p. 2529, 2019

    work page 2019

  10. [10]

    On mathews correlation coefficient and improved distance map loss for automatic glacier calving front segmentation in sar imagery,

    A. Davari, S. Islam, T. Seehaus, A. Hartmann, M. Braun, A. Maier, and V . Christlein, “On mathews correlation coefficient and improved distance map loss for automatic glacier calving front segmentation in sar imagery,” IEEE Transactions on Geoscience and Remote Sensing , vol. 60, pp. 1–12, 2021

    work page 2021

  11. [11]

    Pixelwise distance regression for glacier calving front detection and segmentation,

    A. Davari, C. Baller, T. Seehaus, M. Braun, A. Maier, and V . Christlein, “Pixelwise distance regression for glacier calving front detection and segmentation,” IEEE Transactions on Geoscience and Remote Sensing , vol. 60, pp. 1–10, 2022

    work page 2022

  12. [12]

    Calving fronts and where to find them: a benchmark dataset and methodology for automatic glacier calving front extraction from synthetic aperture radar imagery,

    N. Gourmelon, T. Seehaus, M. Braun, A. Maier, and V . Christlein, “Calving fronts and where to find them: a benchmark dataset and methodology for automatic glacier calving front extraction from synthetic aperture radar imagery,” Earth System Science Data, vol. 14, no. 9, pp. 4287–4313, 2022. [Online]. Available: https://essd.copernicus.org/articles/14/4287/2022/

    work page 2022

  13. [13]

    Conditional random fields for improving deep learning- based glacier calving front delineations,

    N. Gourmelon, J. Klink, T. Seehaus, M. Braun, A. Maier, and V . Christlein, “Conditional random fields for improving deep learning- based glacier calving front delineations,” in IEEE International Geo- science and Remote Sensing Symposium (IGARSS) , 2023, pp. 4939– 4942

    work page 2023

  14. [14]

    A Full-Parameters Microwave Properties Measurement System of 20m Diameter Anechoic Chamber,

    A. Hartmann, A. Davari, T. Seehaus, M. Braun, A. Maier, and V . Christlein, “Bayesian u-net for segmenting glaciers in sar imagery,” in IEEE International Geoscience and Remote Sensing Symposium (IGARSS) , 2021, pp. 3479–3482. [Online]. Available: 10.1109/IGARSS47720.2021.9554292

    work page doi:10.1109/igarss47720.2021.9554292 2021

  15. [15]

    Hed- unet: Combined segmentation and edge detection for monitoring the antarctic coastline,

    K. Heidler, L. Mou, C. Baumhoer, A. Dietz, and X. X. Zhu, “Hed- unet: Combined segmentation and edge detection for monitoring the antarctic coastline,” IEEE Transactions on Geoscience and Remote Sensing, vol. 60, pp. 1–14, 2021

    work page 2021

  16. [16]

    Out-of-the-box calving-front detection method using deep learning,

    O. Herrmann, N. Gourmelon, T. Seehaus, A. Maier, J. J. F ¨urst, M. H. Braun, and V . Christlein, “Out-of-the-box calving-front detection method using deep learning,” The Cryosphere , vol. 17, no. 11, pp. 4957–4977, 2023. [Online]. Available: https://tc.copernicus.org/articles/ 17/4957/2023/

    work page 2023

  17. [17]

    A Full-Parameters Microwave Properties Measurement System of 20m Diameter Anechoic Chamber,

    M. Holzmann, A. Davari, T. Seehaus, M. Braun, A. Maier, and V . Christlein, “Glacier calving front segmentation using attention u-net,” in IEEE International Geoscience and Remote Sensing Symposium (IGARSS) , 2021, pp. 3483–3486. [Online]. Available: 10.1109/IGARSS47720.2021.9555067

    work page doi:10.1109/igarss47720.2021.9555067 2021

  18. [18]

    Extracting glacier calving fronts by deep learning: the benefit of multi-spectral, topographic and textural input features,

    E. Loebel, M. Scheinert, M. Horwath, K. Heidler, J. Christmann, L. D. Phan, A. Humbert, and X. X. Zhu, “Extracting glacier calving fronts by deep learning: the benefit of multi-spectral, topographic and textural input features,” IEEE Transactions on Geoscience and Remote Sensing , vol. 60, pp. 1–12, 2022

    work page 2022

  19. [19]

    Detection of glacier calving margins with convolutional neural networks: A case study,

    Y . Mohajerani, M. Wood, I. Velicogna, and E. Rignot, “Detection of glacier calving margins with convolutional neural networks: A case study,” Remote Sensing, vol. 11, no. 1, p. 74, 2019

    work page 2019

  20. [20]

    How to get the most out of u-net for glacier calving front segmentation,

    M. Periyasamy, A. Davari, T. Seehaus, M. Braun, A. Maier, and V . Christlein, “How to get the most out of u-net for glacier calving front segmentation,” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing , vol. 15, pp. 1712–1723, 2022

    work page 2022

  21. [21]

    Amd-hooknet for glacier front segmentation,

    F. Wu, N. Gourmelon, T. Seehaus, J. Zhang, M. Braun, A. Maier, and V . Christlein, “Amd-hooknet for glacier front segmentation,” IEEE Transactions on Geoscience and Remote Sensing , vol. 61, pp. 1–12, 2023

    work page 2023

  22. [22]

    Automatically delineating the calving front of jakobshavn isbræ from multitemporal terrasar-x images: a deep learning approach,

    E. Zhang, L. Liu, and L. Huang, “Automatically delineating the calving front of jakobshavn isbræ from multitemporal terrasar-x images: a deep learning approach,” The Cryosphere , vol. 13, no. 6, pp. 1729–1741, 2019

    work page 2019

  23. [23]

    U-net: Convolutional networks for biomedical image segmentation,

    O. Ronneberger, P. Fischer, and T. Brox, “U-net: Convolutional networks for biomedical image segmentation,” in Medical Image Computing and Computer-Assisted Intervention (MICCAI) , N. Navab, J. Hornegger, W. M. Wells, and A. F. Frangi, Eds. Cham: Springer International Publishing, 2015, pp. 234–241

    work page 2015

  24. [24]

    Calving front machine (calfin): glacial termini dataset and automated deep learning extraction method for greenland, 1972–2019,

    D. Cheng, W. Hayes, E. Larour, Y . Mohajerani, M. Wood, I. Velicogna, and E. Rignot, “Calving front machine (calfin): glacial termini dataset and automated deep learning extraction method for greenland, 1972–2019,” The Cryosphere , vol. 15, no. 3, pp. 1663–1675, 2021. [Online]. Available: https://tc.copernicus.org/articles/15/1663/2021/

    work page 1972

  25. [25]

    A deep active contour model for delineating glacier calving fronts,

    K. Heidler, L. Mou, E. Loebel, M. Scheinert, S. Lef `evre, and X. X. Zhu, “A deep active contour model for delineating glacier calving fronts,” IEEE Transactions on Geoscience and Remote Sensing , vol. 61, pp. 1– 12, 2023

    work page 2023

  26. [26]

    Image classification of marine-terminating outlet glaciers in greenland using deep learning methods,

    M. Marochov, C. R. Stokes, and P. E. Carbonneau, “Image classification of marine-terminating outlet glaciers in greenland using deep learning methods,” The Cryosphere , vol. 15, no. 11, pp. 5041–5059, 2021. [Online]. Available: https://tc.copernicus.org/articles/15/5041/2021/ 10

    work page 2021

  27. [27]

    An automated, general- ized, deep-learning-based method for delineating the calving fronts of greenland glaciers from multi-sensor remote sensing imagery,

    E. Zhang, L. Liu, L. Huang, and K. S. Ng, “An automated, general- ized, deep-learning-based method for delineating the calving fronts of greenland glaciers from multi-sensor remote sensing imagery,” Remote Sensing of Environment, vol. 254, p. 112265, 2021

    work page 2021

  28. [28]

    Autoterm: an automated pipeline for glacier terminus extraction using machine learning and a “big data

    E. Zhang, G. Catania, and D. T. Trugman, “Autoterm: an automated pipeline for glacier terminus extraction using machine learning and a “big data” repository of greenland glacier termini,” The Cryosphere, vol. 17, no. 8, pp. 3485–3503, 2023. [Online]. Available: https://tc.copernicus.org/articles/17/3485/2023/

    work page 2023

  29. [29]

    Encoder- decoder with atrous separable convolution for semantic image segmen- tation,

    L.-C. Chen, Y . Zhu, G. Papandreou, F. Schroff, and H. Adam, “Encoder- decoder with atrous separable convolution for semantic image segmen- tation,” in Proceedings of the European Conference on Computer Vision (ECCV), 09 2018

    work page 2018

  30. [30]

    Xception: Deep learning with depthwise separable convolu- tions,

    F. Chollet, “Xception: Deep learning with depthwise separable convolu- tions,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), July 2017

    work page 2017

  31. [31]

    Very Deep Convolutional Networks for Large-Scale Image Recognition

    K. Simonyan and A. Zisserman, “Very deep convolutional networks for large-scale image recognition,” in International Conference on Learning Representations (ICLR), Y . Bengio and Y . LeCun, Eds., 2015. [Online]. Available: http://arxiv.org/abs/1409.1556

    work page internal anchor Pith review Pith/arXiv arXiv 2015

  32. [32]

    Gla- stdeeplab: Sar enhancing glacier and ice shelf front detection using swin-transdeeplab with global–local attention,

    Q. Zhu, H. Guo, L. Zhang, D. Liang, Z. Wu, Y . Liu, and Z. Lv, “Gla- stdeeplab: Sar enhancing glacier and ice shelf front detection using swin-transdeeplab with global–local attention,” IEEE Transactions on Geoscience and Remote Sensing , vol. 61, pp. 1–13, 2023

    work page 2023

  33. [33]

    An image is worth 16x16 words: Transformers for image recognition at scale,

    A. Dosovitskiy, L. Beyer, A. Kolesnikov, D. Weissenborn, X. Zhai, T. Unterthiner, M. Dehghani, M. Minderer, G. Heigold, S. Gelly, J. Uszkoreit, and N. Houlsby, “An image is worth 16x16 words: Transformers for image recognition at scale,” in International Conference on Learning Representations , 2021. [Online]. Available: https://openreview.net/forum?id=YicbFdNTTy

    work page 2021

  34. [34]

    Contextual hookformer for glacier calving front segmentation,

    F. Wu, N. Gourmelon, T. Seehaus, J. Zhang, M. Braun, A. Maier, and V . Christlein, “Contextual hookformer for glacier calving front segmentation,” IEEE Transactions on Geoscience and Remote Sensing , vol. 62, pp. 1–15, 2024

    work page 2024

  35. [35]

    On the Opportunities and Risks of Foundation Models

    R. Bommasani, D. A. Hudson, E. Adeli, R. Altman, S. Arora, S. von Arx, M. S. Bernstein, J. Bohg, A. Bosselut, E. Brunskill, E. Brynjolfsson, S. Buch, D. Card, R. Castellon, N. Chatterji, A. Chen, K. Creel, J. Q. Davis, D. Demszky, C. Donahue, M. Doumbouya, E. Durmus, S. Ermon, J. Etchemendy, K. Ethayarajh, L. Fei-Fei, C. Finn, T. Gale, L. Gille- spie, K. ...

    work page internal anchor Pith review Pith/arXiv arXiv 2022

  36. [36]

    Segment anything,

    A. Kirillov, E. Mintun, N. Ravi, H. Mao, C. Rolland, L. Gustafson, T. Xiao, S. Whitehead, A. C. Berg, L. Wan-Yen, P. Doll ´ar, and R. Girshick, “Segment anything,” in Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , October 2023, pp. 4015–4026

    work page 2023

  37. [37]

    Seggpt: Segmenting everything in context,

    X. Wang, X. Zhang, Y . Cao, W. Wang, C. Shen, and T. Huang, “Seggpt: Segmenting everything in context,” 2023, preprint at https://arxiv.org/abs/2304.03284. [Online]. Available: https://arxiv.org/ abs/2304.03284

    work page arXiv 2023

  38. [38]

    Segment everything everywhere all at once,

    X. Zou, J. Yang, H. Zhang, F. Li, L. Li, J. Gao, and Y . J. Lee, “Segment everything everywhere all at once,” in Advances in Neural Information Processing Systems , A. Oh, T. Naumann, A. Globerson, K. Saenko, M. Hardt, and S. Levine, Eds., vol. 36. Curran Associates, Inc., 2023, pp. 19 769–19 782. [Online]. Available: https://proceedings.neurips.cc/paper ...

    work page 2023

  39. [39]

    Family of boundary overlap metrics for the evaluation of medical image segmentation,

    V . Yeghiazaryan and I. V oiculescu, “Family of boundary overlap metrics for the evaluation of medical image segmentation,” Journal of Medical Imaging, vol. 5, no. 1, p. 015006, Feb. 2018

    work page 2018

  40. [40]

    Les distances de chanfrein en analyse d’images : fondements et applications,

    E. Thiel, “Les distances de chanfrein en analyse d’images : fondements et applications,” Theses, Universit ´e Joseph-Fourier - Grenoble I, Sep

    work page

  41. [41]

    Available: https://theses.hal.science/tel-00005113

    [Online]. Available: https://theses.hal.science/tel-00005113

    work page

  42. [42]

    Use of ranks in one-criterion variance analysis,

    H. Kruskal, W. and W. Wallis, W. “Use of ranks in one-criterion variance analysis,” Journal of the American Statistical Association , vol. 47, pp. 583–621, 1952

    work page 1952

  43. [43]

    Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected crfs,

    L.-C. Chen, G. Papandreou, I. Kokkinos, K. Murphy, and A. L. Yuille, “Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected crfs,” IEEE T. Pattern. Anal. , vol. 40, no. 4, pp. 834–848, 2018

    work page 2018

  44. [44]

    Hooknet: Multi-resolution convolutional neural networks for semantic segmentation in histopathology whole-slide images,

    M. van Rijthoven, M. Balkenhol, K. Silin ¸a, J. van der Laak, and F. Ciompi, “Hooknet: Multi-resolution convolutional neural networks for semantic segmentation in histopathology whole-slide images,” Medical Image Analysis , vol. 68, p. 101890, 2021. [Online]. Available: https://www.sciencedirect.com/science/article/pii/S1361841520302541

    work page 2021

  45. [45]

    The unquantified mass loss of northern hemisphere marine-terminating glaciers from 2000–2020,

    W. Kochtitzky, L. Copland, W. van Wychen, R. Hugonnet, R. Hock, J. A. Dowdeswell, T. Benham, T. Strozzi, A. Glazovsky, I. Lavrentiev, D. R. Rounce, R. Millan, A. Cook, A. Dalton, H. Jiskoot, J. Cooley, J. Jania, and F. Navarro, “The unquantified mass loss of northern hemisphere marine-terminating glaciers from 2000–2020,” Nature communications, vol. 13, n...

    work page 2000

  46. [46]

    Progress toward globally complete frontal ablation estimates of marine-terminating glaciers,

    W. Kochtitzky, L. Copland, W. Van Wychen, R. Hock, D. R. Rounce, H. Jiskoot, T. A. Scambos, M. Morlighem, M. King, L. Cha, and et al., “Progress toward globally complete frontal ablation estimates of marine-terminating glaciers,” Annals of Glaciology , vol. 63, no. 87–89, p. 143–152, 2022

    work page 2022

  47. [47]

    CaFFe (CAlving Fronts and where to Find thEm: a benchmark dataset and methodology for automatic glacier calving front extraction from sar imagery),

    N. Gourmelon, T. Seehaus, M. H. Braun, A. Maier, and V . Christlein, “CaFFe (CAlving Fronts and where to Find thEm: a benchmark dataset and methodology for automatic glacier calving front extraction from sar imagery),” 2022, PANGAEA, https://doi.org/10.1594/PANGAEA.940950. [Online]. Available: https://doi.org/10.1594/PANGAEA.940950

    work page doi:10.1594/pangaea.940950 2022

  48. [48]

    Calfin: Calving front dataset for east/west greenland, 1972-2019,

    D. Cheng, W. Hayes, and E. Larour, “Calfin: Calving front dataset for east/west greenland, 1972-2019,” 2020, Dryad, https://doi.org/10.7280/D1FH5D

    work page doi:10.7280/d1fh5d 1972

  49. [49]

    Generalised dice overlap as a deep learning loss function for highly unbalanced segmentations,

    C. H. Sudre, W. Li, T. Vercauteren, S. Ourselin, and M. Jorge Cardoso, “Generalised dice overlap as a deep learning loss function for highly unbalanced segmentations,” inDeep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support , 2017, pp. 240– 248

    work page 2017

  50. [50]

    C. M. Bishop, Neural networks for pattern recognition, 14th ed. Claren- don Press, 1995

    work page 1995

  51. [51]

    nnu-net: a self-configuring method for deep learning-based biomedical image segmentation,

    F. Isensee, P. F. Jaeger, S. A. A. Kohl, J. Petersen, and K. H. Maier-Hein, “nnu-net: a self-configuring method for deep learning-based biomedical image segmentation,”Nature methods, vol. 18, no. 2, pp. 203–211, 2021

    work page 2021

  52. [52]

    Swin transformer: Hierarchical vision transformer using shifted windows,

    Z. Liu, Y . Lin, Y . Cao, H. Hu, Y . Wei, Z. Zhang, S. Lin, and B. Guo, “Swin transformer: Hierarchical vision transformer using shifted windows,” inProceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) , October 2021, pp. 10 012–10 022

    work page 2021

  53. [53]

    Manually delineated calving fronts at Jakobshavn Isbræ, Kangerlussuaq, and Helheim,

    E. Zhang, L. Liu, L. Huang, and K. S. Ng, “Manually delineated calving fronts at Jakobshavn Isbræ, Kangerlussuaq, and Helheim,” 2020, PANGAEA, https://doi.org/10.1594/PANGAEA.923270

    work page doi:10.1594/pangaea.923270 2020

  54. [54]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” inProceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , June 2016

    work page 2016

  55. [55]

    Dilated residual networks,

    F. Yu, V . Koltun, and T. Funkhouser, “Dilated residual networks,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), July 2017

    work page 2017

  56. [56]

    MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications

    A. G. Howard, M. Zhu, B. Chen, D. Kalenichenko, W. Wang, T. Weyand, M. Andreetto, and H. Adam, “Mobilenets: Efficient convolutional neural networks for mobile vision applications,” 2017, preprint at http://arxiv.org/abs/1704.04861

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  57. [57]

    Termpicks: a century of greenland glacier terminus data for use in scientific and machine learning applications,

    S. Goliber, T. Black, G. Catania, J. M. Lea, H. Olsen, D. Cheng, S. Bevan, A. Bjørk, C. Bunce, S. Brough, J. R. Carr, T. Cowton, A. Gardner, D. Fahrner, E. Hill, I. Joughin, N. J. Korsgaard, A. Luckman, T. Moon, T. Murray, A. Sole, M. Wood, and E. Zhang, “Termpicks: a century of greenland glacier terminus data for use in scientific and machine learning ap...

    work page 2022

  58. [58]

    Measuring and Modeling Mountain glaciers and ice caps in a Changing Cli- mAte (M3OCCA)

    He is currently a postdoctoral research fellow with the Pattern Recognition Laboratory, Friedrich- Alexander-Universit¨at Erlangen-N¨urnberg, Erlangen, Germany. His research interests include computer vision and machine learning with a focus on hand- writing analysis, object tracking, and segmentation. Noah Maul received his B.Sc. and M.Sc. de- grees in c...

    work page 2020

  59. [59]

    Measuring and Modeling Mountain glaciers and ice caps in a Changing ClimAte (M3OCCA)

    Later on, he joined the Pattern Recognition Lab in October 2023 as a Ph.D candidate under the supervision of Prof. Andreas Maier. His current research focuses on machine learning on radargrams. Dakota Pyles received a B.Sc. in Geosciences from the University of Montana in 2019 and a M.Sc. in Geology from the University of Idaho in 2022. He is currently pu...

    work page 2023

  60. [60]

    Zones” and “Front

    for calving front extraction. The best-performing version uses Matthew’s Correlation Coefficient (MCC) as an early stopping criterion and is trained on binary segmentation masks showing the calving front versus background using an im- proved distance map loss. Davari et al. [10]’s dataset includes SAR imagery of the Jakobshavn Isbrae Glacier located in Gr...

    work page 2017

  61. [61]

    Parallel; [13]; [36] Iterative; [15] Zones) and sometimes no ocean is predicted at all ( [32]; [36] Parallel; [36] Iterative). When the ocean is predicted in the correct location of the image, the ocean outline and, thus, the calving front often do not have the correct shape ( [20]; [14]; [32]; [28]; [12] Zones,

    work page

  62. [62]

    In binary front segmentation, the predicted fronts in the majority of images only cover parts of the ground truth, and many additional false positive fronts are predicted [12]

    Parallel; [13]; [36] Iterative; [15] Zones). In binary front segmentation, the predicted fronts in the majority of images only cover parts of the ground truth, and many additional false positive fronts are predicted [12]. Only five systems have an MDE lower than 600 m: Loebel et al. [18], Herrmann et al. [16], Heidler et al. [15]’s front output, Wu et al....

    work page