Pith. sign in

REVIEW 2 major objections 1 minor 56 references

PolyBuild extracts vector building polygons directly from high-resolution remote sensing images in one forward pass without post-processing.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · grok-4.3

2026-06-27 17:22 UTC pith:YA4EIFTF

load-bearing objection PolyBuild sketches a direct polygon pipeline via sub-region centers and CNN-Transformer refinement, but the abstract supplies no metrics or output-format details to support the no-post-processing claim. the 2 major comments →

arxiv 2606.08920 v1 pith:YA4EIFTF submitted 2026-06-08 cs.CV cs.AI

PolyBuild: An End-to-End Method for Polygonal Building Contour Extraction from High-Resolution Remote Sensing Images

Yaoteng Zhang , Julin Zhang , Guangshuai Wang , Jiwei Deng , Hui Sheng , Yasir Muhammad , Shiqing Wei This is my paper
classification cs.CV cs.AI
keywords building polygon extractionremote sensing imagesend-to-end contour extractionCNN-Transformer hybridvector polygon outputinitial contour generationhigh-resolution imagerycontour optimization
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The paper introduces PolyBuild, an end-to-end neural network that produces ready-to-use building polygons from remote sensing images. It relies on an Initial Contour Generation Module that represents each building with center features from four sub-regions and a Contour Optimization Module that refines those shapes. The refinement step combines CNN local features with Transformer global positional encoding in an iterative decoder. This design targets the problems of error-prone multi-step pipelines that first segment pixels and then convert to vectors. If the approach holds, it would let mapping systems output precise vector data straight from imagery for applications like urban planning.

Core claim

PolyBuild directly extracts building vector polygons from high-resolution remote sensing images using an Initial Contour Generation Module to produce starting contours from concatenated sub-region center features and a Contour Optimization Module that iteratively refines them by integrating CNN features with contour positional information inside a Transformer-based decoder, delivering higher performance than prior mask-based and contour-based methods on three building datasets without any post-processing operations.

What carries the argument

The hybrid CNN-Transformer decoder in the Contour Optimization Module, which refines initial contours by repeatedly merging local image features with global contour position data.

Load-bearing premise

The hybrid CNN-Transformer decoder can reliably refine initial contours to high boundary accuracy under varying imaging conditions and complex building shapes.

What would settle it

On the three evaluation datasets, PolyBuild polygon outputs show lower boundary accuracy or higher vertex error than a standard segmentation model followed by standard vectorization post-processing.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Building extraction reduces to a single network forward pass instead of a pipeline of segmentation, vectorization, and cleanup stages.
  • Error accumulation from separate post-processing steps is removed, which can improve consistency of polygon boundaries.
  • Direct vector outputs become available for immediate use in geographic information systems without format conversion.
  • The method handles simultaneous detection and contouring for multiple building instances in one image.

Where Pith is reading between the lines

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

  • The two-module design could extend to direct polygonal extraction of other linear or areal features such as roads or field boundaries from the same imagery.
  • Integration with multi-date image stacks might support change detection of building footprints over time with minimal additional processing.
  • If the refinement step generalizes, similar hybrid decoders could be tested on lower-resolution or multispectral sensors.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

2 major / 1 minor

Summary. The paper proposes PolyBuild, an end-to-end framework for extracting building polygons from high-resolution remote sensing images. It uses an Initial Contour Generation Module (ICGM) to produce initial contours via bounding boxes and sub-region center features, followed by a Contour Optimization Module (COM) that iteratively refines contours with a hybrid CNN-Transformer decoder. The central claims are that the method outputs ready-to-use vector polygons with no post-processing and significantly outperforms prior mask-based and contour-based methods on three datasets.

Significance. If the performance claims hold with rigorous evidence, the work would advance automated mapping by removing post-processing stages that introduce errors and overhead. The ICGM's use of concatenated sub-region centers for simultaneous detection and contour initialization is a concrete design contribution that could be reusable.

major comments (2)
  1. [Abstract; Contour Optimization Module] Abstract and Contour Optimization Module description: the assertion that the method 'directly extract[s] building vector polygons ... without the need for any post-processing operations' is load-bearing for the title and abstract claims, yet the decoder output representation is not specified (e.g., whether the final COM iteration emits ordered vertex lists or still requires polygonization, simplification, or snapping).
  2. [Abstract] Abstract: the claim that PolyBuild 'significantly outperforms state-of-the-art methods' on three datasets is unsupported by any quantitative metrics, tables, ablation studies, or error analysis in the manuscript text, so the central empirical claim cannot be evaluated.
minor comments (1)
  1. [Contour Optimization Module] The hybrid CNN-Transformer description would benefit from a diagram or pseudocode showing the exact integration of CNN features with contour positional encodings across iterations.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback. We address each major comment below and indicate where revisions will be made to the manuscript.

read point-by-point responses

  1. Referee: [Abstract; Contour Optimization Module] Abstract and Contour Optimization Module description: the assertion that the method 'directly extract[s] building vector polygons ... without the need for any post-processing operations' is load-bearing for the title and abstract claims, yet the decoder output representation is not specified (e.g., whether the final COM iteration emits ordered vertex lists or still requires polygonization, simplification, or snapping).

    Authors: We agree that the precise output representation of the final COM iteration must be stated explicitly to support the no-post-processing claim. The COM decoder is intended to emit an ordered sequence of vertex coordinates that directly define the closed polygon. We will revise both the abstract and the Contour Optimization Module section to specify that the output is an ordered vertex list forming a vector polygon, with no subsequent polygonization, simplification, or snapping required. revision: yes

  2. Referee: [Abstract] Abstract: the claim that PolyBuild 'significantly outperforms state-of-the-art methods' on three datasets is unsupported by any quantitative metrics, tables, ablation studies, or error analysis in the manuscript text, so the central empirical claim cannot be evaluated.

    Authors: The referee correctly notes that the abstract makes a strong performance claim without quantitative support in the text. Although the manuscript contains experimental results and comparisons on the three datasets, these are not summarized or referenced within the abstract itself. We will revise the abstract to include a concise statement of key quantitative improvements drawn from the experimental section, thereby making the claim evaluable directly from the abstract. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical architecture with no derivations or fitted predictions

full rationale

The paper describes an empirical CNN-Transformer architecture (ICGM + iterative COM) for direct polygon extraction and reports dataset performance; no equations, parameter fits, or self-citation chains are invoked to derive the central claim. The 'end-to-end without post-processing' assertion is an empirical design claim evaluated on benchmarks rather than a reduction to prior fitted quantities by construction. No load-bearing steps match any enumerated circularity pattern.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract supplies no explicit free parameters, mathematical axioms, or new physical entities; the two named modules are architectural components rather than postulated entities with independent evidence.

pith-pipeline@v0.9.1-grok · 5813 in / 1037 out tokens · 18267 ms · 2026-06-27T17:22:55.626292+00:00 · methodology

0 comments
read the original abstract

Extracting building polygon contours from high-resolution remote sensing images is a fundamental task for various mapping applications. However, the presence of varying imaging conditions and complex building structures, makes automatic contour extraction extremely challenging. Mainstream approaches for building extraction often rely on pixel-level segmentation followed by multiple post-processing steps to produce building contour, which can be computationally intensive and prone to errors. In this paper, we propose an end-to-end method named PolyBuild, which can directly extract building vector polygons from high-resolution remote sensing images without the need for any post-processing operations. The proposed method leverages two primary modules: an Initial Contour Generation Module (ICGM) and a Contour Optimization Module (COM). The ICGM is designed to generate an initial building contour by utilizing concatenated sub-region center features for each building instance. It performs simultaneous object detection and initial contour extraction by generating bounding boxes and using the center features of four sub-regions to represent each building. The Contour Optimization Module (COM) further refines the generated building contours by iteratively integrating Convolutional Neural Network (CNN) features and contour positional information in a Transformer-based decoder. The hybrid CNN-Transformer architecture effectively captures both local and global spatial relationships within the building contour, ensuring high-quality boundary delineation. Extensive experiments are conducted on three building datasets to evaluate the performance of PolyBuild. The results demonstrate that PolyBuild significantly outperforms state-of-the-art methods, including mask-based and contour-based approaches.

Figures

Figures reproduced from arXiv: 2606.08920 by Guangshuai Wang, Hui Sheng, Jiwei Deng, Julin Zhang, Shiqing Wei, Yaoteng Zhang, Yasir Muhammad.

Figure 1
Figure 1. Figure 1: Different methods generate initial contours as follows: (a) DANCE [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overall architecture of the PolyBuild. PolyBuild generates the initial contour through Initial Contour Generation Moduel (ICGM) and then inputs the [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: The architecture of the Initial Contour Generation Module (ICGM). First, the ICGM predicts a bbox for each building instance and divides the bbox [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Comparison of building contour vertex offset regression between [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: The architecture of contour optimization module integrated CNN-Transformer. [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Example of dynamic matching loss. (a) A predicted contour vertex is [PITH_FULL_IMAGE:figures/full_fig_p007_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Visualization results of different methods on the WHU aerial building dataset. [PITH_FULL_IMAGE:figures/full_fig_p009_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Visualization results of different methods on the WHU-Mix building dataset. [PITH_FULL_IMAGE:figures/full_fig_p010_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Visualization of initial contour generation methods under different [PITH_FULL_IMAGE:figures/full_fig_p011_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Examples of prediction results of PolyBuild and YOLACT under [PITH_FULL_IMAGE:figures/full_fig_p012_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Qualitative results obtained by PolyBuild. From top to bottom: results on the WHU aerial building dataset, WHU-Mix test set I, WHU-Mix test set [PITH_FULL_IMAGE:figures/full_fig_p013_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Examples of buildings that PolyBuild fails to extract correctly. (a) [PITH_FULL_IMAGE:figures/full_fig_p013_12.png] view at source ↗

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

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

  1. [1]

    Polyworld: Polygonal building extraction with graph neural networks in satellite images,

    S. Zorzi, S. Bazrafkan, S. Habenschuss, and F. Fraundorfer, “Polyworld: Polygonal building extraction with graph neural networks in satellite images,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 1848–1857

    2022

  2. [2]

    Gated spatial memory and centroid- aware network for building instance extraction,

    L. Xu, Y . Li, J. Xu, and L. Guo, “Gated spatial memory and centroid- aware network for building instance extraction,”IEEE Transactions on Geoscience and Remote Sensing, vol. 60, pp. 1–14, 2021

    2021

  3. [3]

    Extracting man-made objects from high spatial resolution remote sensing images via fast level set evolutions,

    Z. Li, W. Shi, Q. Wang, and Z. Miao, “Extracting man-made objects from high spatial resolution remote sensing images via fast level set evolutions,”IEEE Transactions on Geoscience and Remote Sensing, vol. 53, no. 2, pp. 883–899, 2014

    2014

  4. [4]

    Scene-driven multitask parallel attention network for building extraction in high- resolution remote sensing images,

    H. Guo, Q. Shi, B. Du, L. Zhang, D. Wang, and H. Ding, “Scene-driven multitask parallel attention network for building extraction in high- resolution remote sensing images,”IEEE Transactions on Geoscience and Remote Sensing, vol. 59, no. 5, pp. 4287–4306, 2020

    2020

  5. [5]

    P. E. Hart, D. G. Stork, R. O. Dudaet al.,Pattern classification. Wiley Hoboken, 2000

    2000

  6. [6]

    Automatic building extraction based on multiresolution segmentation using remote sensing data,

    N. Shrivastava and P. Kumar Rai, “Automatic building extraction based on multiresolution segmentation using remote sensing data,”Geographia Polonica, vol. 88, no. 3, pp. 407–421, 2015

    2015

  7. [7]

    A computational approach to edge detection,

    J. Canny, “A computational approach to edge detection,”IEEE Transac- tions on pattern analysis and machine intelligence, no. 6, pp. 679–698, 1986

    1986

  8. [8]

    The extraction of building shadow and the estimation of building heights based on morphology and spectral characteristic parameters,

    H. Yunfeng and Z. Qianli, “The extraction of building shadow and the estimation of building heights based on morphology and spectral characteristic parameters,”Bulletin of Surveying and Mapping, no. 6, p. 22, 2018. 13 Fig. 11. Qualitative results obtained by PolyBuild. From top to bottom: results on the WHU aerial building dataset, WHU-Mix test set I, WHU...

    2018

  9. [9]

    A multidirectional and multiscale morpholog- ical index for automatic building extraction from multispectral geoeye- 1 imagery,

    X. Huang and L. Zhang, “A multidirectional and multiscale morpholog- ical index for automatic building extraction from multispectral geoeye- 1 imagery,”Photogrammetric Engineering & Remote Sensing, vol. 77, no. 7, pp. 721–732, 2011

    2011

  10. [10]

    Automatic 3d building reconstruction from multi-view aerial images with deep learning,

    D. Yu, S. Ji, J. Liu, and S. Wei, “Automatic 3d building reconstruction from multi-view aerial images with deep learning,”ISPRS Journal of Photogrammetry and Remote Sensing, vol. 171, pp. 155–170, 2021

    2021

  11. [11]

    Decoupling semantic and edge representations for building footprint extraction from remote sensing images,

    H. Guo, X. Su, C. Wu, B. Du, and L. Zhang, “Decoupling semantic and edge representations for building footprint extraction from remote sensing images,”IEEE Transactions on Geoscience and Remote Sensing, vol. 61, pp. 1–16, 2023

    2023

  12. [12]

    Shipgeonet: Sar image-based geometric feature extraction of ships using convolutional neural networks,

    M. Yasir, S. Liu, X. Mingming, J. Wan, S. Pirasteh, and K. B. Dang, “Shipgeonet: Sar image-based geometric feature extraction of ships using convolutional neural networks,”IEEE Transactions on Geoscience and Remote Sensing, 2024

    2024

  13. [13]

    Building extraction from high-resolution multispectral and sar images using a boundary- link multimodal fusion network,

    Z. Zhao, B. Zhao, Y . Wu, Z. He, and L. Gao, “Building extraction from high-resolution multispectral and sar images using a boundary- link multimodal fusion network,”IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, pp. 1–15, 2025

    2025

  14. [14]

    Toward automatic building footprint delineation from aerial images using cnn and regularization,

    S. Wei, S. Ji, and M. Lu, “Toward automatic building footprint delineation from aerial images using cnn and regularization,”IEEE Transactions on Geoscience and Remote Sensing, vol. 58, no. 3, pp. 2178–2189, 2019

    2019

  15. [15]

    Bomsc- net: Boundary optimization and multi-scale context awareness based building extraction from high-resolution remote sensing imagery,

    Y . Zhou, Z. Chen, B. Wang, S. Li, H. Liu, D. Xu, and C. Ma, “Bomsc- net: Boundary optimization and multi-scale context awareness based building extraction from high-resolution remote sensing imagery,”IEEE Transactions on Geoscience and Remote Sensing, vol. 60, pp. 1–17, 2022

    2022

  16. [16]

    Map-net: Multiple attending path neural network for building footprint extraction from remote sensed imagery,

    Q. Zhu, C. Liao, H. Hu, X. Mei, and H. Li, “Map-net: Multiple attending path neural network for building footprint extraction from remote sensed imagery,”IEEE Transactions on Geoscience and Remote Sensing, vol. 59, no. 7, pp. 6169–6181, 2020. 14

    2020

  17. [17]

    Boundary shape- preserving model for building mapping from high-resolution remote sensing images,

    A. Hu, L. Wu, S. Chen, Y . Xu, H. Wang, and Z. Xie, “Boundary shape- preserving model for building mapping from high-resolution remote sensing images,”IEEE Transactions on Geoscience and Remote Sensing, vol. 61, pp. 1–17, 2023

    2023

  18. [18]

    Mask r-cnn,

    K. He, G. Gkioxari, P. Doll ´ar, and R. Girshick, “Mask r-cnn,” in Proceedings of the IEEE international conference on computer vision, 2017, pp. 2961–2969

    2017

  19. [19]

    Yolact: Real-time instance segmentation,

    D. Bolya, C. Zhou, F. Xiao, and Y . J. Lee, “Yolact: Real-time instance segmentation,” inProceedings of the IEEE/CVF international conference on computer vision, 2019, pp. 9157–9166

    2019

  20. [20]

    Polarmask: Single shot instance segmentation with polar rep- resentation,

    E. Xie, P. Sun, X. Song, W. Wang, X. Liu, D. Liang, C. Shen, and P. Luo, “Polarmask: Single shot instance segmentation with polar rep- resentation,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 12 193–12 202

    2020

  21. [21]

    Deep snake for real-time instance segmentation,

    S. Peng, W. Jiang, H. Pi, X. Li, H. Bao, and X. Zhou, “Deep snake for real-time instance segmentation,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 8533– 8542

    2020

  22. [22]

    Dance: A deep attentive contour model for efficient instance segmentation,

    Z. Liu, J. H. Liew, X. Chen, and J. Feng, “Dance: A deep attentive contour model for efficient instance segmentation,” inProceedings of the IEEE/CVF winter conference on applications of computer vision, 2021, pp. 345–354

    2021

  23. [23]

    E2ec: An end-to-end contour-based method for high-quality high-speed instance segmentation,

    T. Zhang, S. Wei, and S. Ji, “E2ec: An end-to-end contour-based method for high-quality high-speed instance segmentation,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 4443–4452

    2022

  24. [24]

    Buildmapper: A fully learnable framework for vectorized building contour extraction,

    S. Wei, T. Zhang, S. Ji, M. Luo, and J. Gong, “Buildmapper: A fully learnable framework for vectorized building contour extraction,”ISPRS Journal of Photogrammetry and Remote Sensing, vol. 197, pp. 87–104, 2023

    2023

  25. [25]

    Fast interactive object annotation with curve-gcn,

    H. Ling, J. Gao, A. Kar, W. Chen, and S. Fidler, “Fast interactive object annotation with curve-gcn,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2019, pp. 5257–5266

    2019

  26. [26]

    U-net: Convolutional networks for biomedical image segmentation,

    O. Ronneberger, P. Fischer, and T. Brox, “U-net: Convolutional networks for biomedical image segmentation,” inMedical image computing and computer-assisted intervention–MICCAI 2015: 18th international con- ference, Munich, Germany, October 5-9, 2015, proceedings, part III 18. Springer, 2015, pp. 234–241

    2015

  27. [27]

    Deep high-resolution repre- sentation learning for human pose estimation,

    K. Sun, B. Xiao, D. Liu, and J. Wang, “Deep high-resolution repre- sentation learning for human pose estimation,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2019, pp. 5693–5703

    2019

  28. [28]

    Boundary enhancement semantic segmentation for building extraction from remote sensed image,

    H. Jung, H.-S. Choi, and M. Kang, “Boundary enhancement semantic segmentation for building extraction from remote sensed image,”IEEE Transactions on Geoscience and Remote Sensing, vol. 60, pp. 1–12, 2021

    2021

  29. [29]

    Bctnet: Bi-branch cross- fusion transformer for building footprint extraction,

    L. Xu, Y . Li, J. Xu, Y . Zhang, and L. Guo, “Bctnet: Bi-branch cross- fusion transformer for building footprint extraction,”IEEE Transactions on Geoscience and Remote Sensing, vol. 61, pp. 1–14, 2023

    2023

  30. [30]

    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. Gellyet al., “An image is worth 16x16 words: Transformers for image recognition at scale,”arXiv preprint arXiv:2010.11929, 2020

    work page internal anchor Pith review Pith/arXiv arXiv 2010

  31. [31]

    Segformer: Simple and efficient design for semantic segmentation with transformers,

    E. Xie, W. Wang, Z. Yu, A. Anandkumar, J. M. Alvarez, and P. Luo, “Segformer: Simple and efficient design for semantic segmentation with transformers,”Advances in neural information processing systems, vol. 34, pp. 12 077–12 090, 2021

    2021

  32. [32]

    Path aggregation network for instance segmentation,

    S. Liu, L. Qi, H. Qin, J. Shi, and J. Jia, “Path aggregation network for instance segmentation,” inProceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 8759–8768

    2018

  33. [33]

    Conditional convolutions for instance segmentation,

    Z. Tian, C. Shen, and H. Chen, “Conditional convolutions for instance segmentation,” inComputer Vision–ECCV 2020: 16th European Con- ference, Glasgow, UK, August 23–28, 2020, Proceedings, Part I 16. Springer, 2020, pp. 282–298

    2020

  34. [34]

    Blendmask: Top-down meets bottom-up for instance segmentation,

    H. Chen, K. Sun, Z. Tian, C. Shen, Y . Huang, and Y . Yan, “Blendmask: Top-down meets bottom-up for instance segmentation,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 8573–8581

    2020

  35. [35]

    Fully convolutional networks for multisource building extraction from an open aerial and satellite imagery data set,

    S. Ji, S. Wei, and M. Lu, “Fully convolutional networks for multisource building extraction from an open aerial and satellite imagery data set,” IEEE Transactions on geoscience and remote sensing, vol. 57, no. 1, pp. 574–586, 2018

    2018

  36. [36]

    Sequentially delineation of rooftops with holes from vhr aerial images using a convolutional recurrent neural network,

    W. Huang, Z. Liu, H. Tang, and J. Ge, “Sequentially delineation of rooftops with holes from vhr aerial images using a convolutional recurrent neural network,”Remote Sensing, vol. 13, no. 21, p. 4271, 2021

    2021

  37. [37]

    Building outline delineation: From aerial images to polygons with an improved end-to-end learning framework,

    W. Zhao, C. Persello, and A. Stein, “Building outline delineation: From aerial images to polygons with an improved end-to-end learning framework,”ISPRS journal of photogrammetry and remote sensing, vol. 175, pp. 119–131, 2021

    2021

  38. [38]

    Building outline delineation from vhr remote sensing images using the convolutional recurrent neural network embedded with line segment information,

    Z. Liu, H. Tang, and W. Huang, “Building outline delineation from vhr remote sensing images using the convolutional recurrent neural network embedded with line segment information,”IEEE Transactions on Geoscience and Remote Sensing, vol. 60, pp. 1–13, 2022

    2022

  39. [39]

    Hisup: Accurate polygonal mapping of buildings in satellite imagery with hierarchical supervision,

    B. Xu, J. Xu, N. Xue, and G.-S. Xia, “Hisup: Accurate polygonal mapping of buildings in satellite imagery with hierarchical supervision,” ISPRS Journal of Photogrammetry and Remote Sensing, vol. 198, pp. 284–296, 2023

    2023

  40. [40]

    Sam- polybuild: Adapting the segment anything model for polygonal building extraction,

    C. Wang, J. Chen, Y . Meng, Y . Deng, K. Li, and Y . Kong, “Sam- polybuild: Adapting the segment anything model for polygonal building extraction,”ISPRS Journal of Photogrammetry and Remote Sensing, vol. 218, pp. 707–720, 2024

    2024

  41. [41]

    Joint semantic–geometric learning for polygonal building segmentation from high-resolution remote sensing images,

    W. Li, W. Zhao, J. Yu, J. Zheng, C. He, H. Fu, and D. Lin, “Joint semantic–geometric learning for polygonal building segmentation from high-resolution remote sensing images,”ISPRS Journal of Photogram- metry and Remote Sensing, vol. 201, pp. 26–37, 2023

    2023

  42. [42]

    From lines to polygons: Polygonal building contour extraction from high-resolution remote sensing imagery,

    S. Wei, T. Zhang, D. Yu, S. Ji, Y . Zhang, and J. Gong, “From lines to polygons: Polygonal building contour extraction from high-resolution remote sensing imagery,”ISPRS Journal of Photogrammetry and Remote Sensing, vol. 209, pp. 213–232, 2024

    2024

  43. [43]

    Focal loss for dense object detection,

    T.-Y . Lin, P. Goyal, R. Girshick, K. He, and P. Doll ´ar, “Focal loss for dense object detection,” inProceedings of the IEEE international conference on computer vision, 2017, pp. 2980–2988

    2017

  44. [44]

    Objects as Points

    X. Zhou, D. Wang, and P. Kr ¨ahenb¨uhl, “Objects as points,”arXiv preprint arXiv:1904.07850, 2019

    work page internal anchor Pith review Pith/arXiv arXiv 1904

  45. [45]

    Crowdai dataset,

    S. P. Mohanty, “Crowdai dataset,” 2018

    2018

  46. [46]

    open ai tanzania building footprint segmentation challenge,

    J. Yap, “open ai tanzania building footprint segmentation challenge,”,” 2018

    2018

  47. [47]

    SpaceNet: A Remote Sensing Dataset and Challenge Series

    A. Van Etten, D. Lindenbaum, and T. M. Bacastow, “Spacenet: A remote sensing dataset and challenge series,”arXiv preprint arXiv:1807.01232, 2018

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  48. [48]

    Can semantic labeling methods generalize to any city? the inria aerial image labeling benchmark,

    E. Maggiori, Y . Tarabalka, G. Charpiat, and P. Alliez, “Can semantic labeling methods generalize to any city? the inria aerial image labeling benchmark,” in2017 IEEE International geoscience and remote sensing symposium (IGARSS). IEEE, 2017, pp. 3226–3229

    2017

  49. [49]

    Solo: Segmenting objects by locations,

    X. Wang, T. Kong, C. Shen, Y . Jiang, and L. Li, “Solo: Segmenting objects by locations,” inComputer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part XVIII

    2020

  50. [50]

    Springer, 2020, pp. 649–665

    2020

  51. [51]

    Rtmdet: An empirical study of designing real-time object detectors,

    C. Lyu, W. Zhang, H. Huang, Y . Zhou, Y . Wang, Y . Liu, S. Zhang, and K. Chen, “Rtmdet: An empirical study of designing real-time object detectors,” 2022

    2022

  52. [52]

    A concentric loop convolutional neural network for manual delineation-level building boundary segmentation from remote-sensing images,

    S. Wei, T. Zhang, and S. Ji, “A concentric loop convolutional neural network for manual delineation-level building boundary segmentation from remote-sensing images,”IEEE Transactions on Geoscience and Remote Sensing, vol. 60, pp. 1–11, 2021

    2021

  53. [53]

    End-to-end object detection with transformers,

    N. Carion, F. Massa, G. Synnaeve, N. Usunier, A. Kirillov, and S. Zagoruyko, “End-to-end object detection with transformers,” in European conference on computer vision. Springer, 2020, pp. 213– 229

    2020

  54. [54]

    Masked-attention mask transformer for universal image segmentation,

    B. Cheng, I. Misra, A. G. Schwing, A. Kirillov, and R. Girdhar, “Masked-attention mask transformer for universal image segmentation,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 1290–1299

    2022

  55. [55]

    Building instance segmentation and boundary regularization from high-resolution remote sensing images,

    W. Zhao, C. Persello, and A. Stein, “Building instance segmentation and boundary regularization from high-resolution remote sensing images,” in IGARSS 2020-2020 IEEE International Geoscience and Remote Sensing Symposium. IEEE, 2020, pp. 3916–3919

    2020

  56. [56]

    Topological map extraction from overhead images,

    Z. Li, J. D. Wegner, and A. Lucchi, “Topological map extraction from overhead images,” inProceedings of the IEEE/CVF International Conference on Computer Vision, 2019, pp. 1715–1724

    2019