pith. sign in

arxiv: 2606.28390 · v1 · pith:QBIRR34Vnew · submitted 2026-06-23 · 💻 cs.CV · cs.AI

Automated Quality Assessment of Geospatial Vector Data: A GeoAI Approach using Spatial Representation Learning

Hao Li , Chen Chu , Filip Biljecki , Cyrus Shahabi , Wenwen Li This is my paper

Pith reviewed 2026-06-30 09:52 UTC · model grok-4.3

classification 💻 cs.CV cs.AI
keywords geospatial vector dataquality assessmentGeoAIspatial representation learningtopological errorsoverlapping polygonsstreet networksautomated detection
0
0 comments X

The pith

Topo4Vec detects topological errors in geospatial vector data by training on simulated examples and isolating faults in a learned latent space.

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

The paper proposes an automated framework called Topo4Vec that replaces manual labeling with simulation of topological errors such as overlapping building polygons and connectivity faults in street networks. It then applies spatial representation learning to map native vector geometries into a latent space where erroneous instances separate from valid ones. Evaluation across Los Angeles, Munich, and Singapore shows the approach reaches 0.99 accuracy on building overlaps and 0.60 on street overshoots and undershoots. The work argues this combination supports scalable, consistent quality monitoring for growing volumes of vector data without repeated human annotation. A sympathetic reader would care because traditional rule-based checks fail on varied urban forms and large datasets, while this method offers a data-driven alternative.

Core claim

Topo4Vec relaxes manual annotation by simulating topological errors such as overlapping polygons and street network overshoots or undershoots, then encodes complex vector geometries with spatial representation learning so that errors become isolated from valid geometries in the resulting latent space.

What carries the argument

Topo4Vec framework, which pairs topological error simulation with spatial representation learning to produce a latent space separating erroneous from valid vector geometries.

If this is right

  • Quality assessment of building footprints and street networks becomes feasible at city scale without per-dataset manual labeling.
  • The same simulation-plus-representation pipeline can be applied to additional error types once simulation rules are defined.
  • Latent-space separation provides a quantitative signal for ranking data layers by consistency across multiple cities.
  • Open release of code and data allows direct replication on new vector datasets from other regions.

Where Pith is reading between the lines

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

  • Integration into live geospatial data pipelines could flag new uploads for review before they enter public maps.
  • Extending the simulation rules to three-dimensional or time-varying vector objects would test whether the latent-space isolation generalizes beyond two-dimensional footprints and networks.

Load-bearing premise

Simulated topological errors generate training examples representative enough that the learned representations can separate real errors from valid geometries.

What would settle it

Running the trained model on a fresh collection of vector data containing only unsimulated, human-verified topological errors and finding accuracy substantially below the reported 0.99 and 0.60 levels.

Figures

Figures reproduced from arXiv: 2606.28390 by Chen Chu, Cyrus Shahabi, Filip Biljecki, Hao Li, Wenwen Li.

Figure 1
Figure 1. Figure 1: A Diagram of Topo4Vec: (a) an Automated GeoAI-based framework for Geospatial Vector Data Quality Assessment, (b) real-world examples of vector data quality issues such as overlapping polygons in OpenStreetMap and inconsistent streets in Microsoft Roads. labor-intensive to identify in map production environments, so the requirement for manual labeling contradicts the very premise of autonomy (Barron et al. … view at source ↗
Figure 2
Figure 2. Figure 2: The technical workflow of Topo4Vec, detailing the data pipeline through three primary phases: (a) data preprocessing, (b) topological error simulation, and (c) neural representation learning for automated vector data quality measurement. native vector data is surprisingly underexplored. Vector data encodes explicit geomet￾ric and topological structures, such as street network layouts and building arrange￾m… view at source ↗
Figure 3
Figure 3. Figure 3: Visual examples of simulated topological errors in building footprints, contrasting valid non￾overlapping structures with simulated overlapping polygons. overlap occurs if their intersection yields a positive area than a strict tolerance ϵ > 0: Eoverlap(pi , pj ) = I [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Visual examples of simulated connectivity anomalies in street networks, distinguishing between simulated undershoot errors, overshoot errors, and correct polyline topologies defined the overlap proportion ρ as the ratio of the intersection area to the area of the smaller polygon: ρ(pi , pj ) = µ [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: The architecture of the Geo2Vec model used in the Topo4Vec, highlighting the neural representation pipeline via Signed Distance Functions (SDF) and the contrastive topological classifier designed to detect overlapping polygons and over- and undershoot street networks. (Oleynikova et al. 2016). In the rest of this section, we will elaborate on the details of SDF approach and the way how it is used in Topo4V… view at source ↗
Figure 6
Figure 6. Figure 6: Three study area selected for evaluating the performance of Topo4Vec, namely Los Angeles (USA), Munich (Germany), and Singapore. other North American cities. LA is characterized by a rectilinear street grid and predominantly orthogonal, single-family residential footprints, interspersed with high-density commercial areas. This topological pattern challenges the evaluation of the model on uniform, convex po… view at source ↗
Figure 7
Figure 7. Figure 7: Selected visualizations of the learned SDF for non-overlapped and overlapped polygon pairs across Los Angeles, Singapore, and Munich. 15 [PITH_FULL_IMAGE:figures/full_fig_p015_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Selected visualization of learned SDF representing overshoot and undershoot errors in street net￾works. Visualization of the overshoot and undershoot SDF maps in [PITH_FULL_IMAGE:figures/full_fig_p017_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Ablation studies comparing contrastive approaches against naive classifiers for overshoot error detection across the three study areas. 3.7. Ablation Study To further validate Topo4Vec’s design and limitation, we designed three ablation stud￾ies. First, we compared contrastive learning approaches against a naive softmax base￾line to examine the actual performance gains from the contrastive learning objecti… view at source ↗
Figure 10
Figure 10. Figure 10: Ablation studies comparing contrastive approaches against naive classifiers for undershoot error detection across the three study areas [PITH_FULL_IMAGE:figures/full_fig_p019_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Train and Test Performance Gap visualized using dumbbell plot to illustrate the generalization robustness of different models for both overshoot and undershoot detection. 4. Discussion This study introduced Topo4Vec, an automated GeoAI framework, designed for topo￾logical quality assessment of vector data (e.g., such as polygons and polylines) using advanced SRL approaches. By evaluating the framework acr… view at source ↗
read the original abstract

Geospatial vector data quality is a foundational research topic in GIS, yet classic rule-based quality assessment algorithms often struggle with diverse urban morphologies and massive data volumes. Recently, Geospatial Artificial Intelligence (GeoAI) shows promising potential for automating geospatial analysis, while its application to native vector data remains largely underexplored. To fill this research gap, we proposed Topo4Vec, an automated GeoAI framework, designed for scalable vector data quality assessment via advanced Spatial Representation Learning (SRL). Specifically, Topo4Vec relax the labor-intensive manual annotation process via topological error simulation, such as overlapping polygons and street network connectivity errors e.g., overshoots and undershoots. Then, it leverages state-of-the-art SRL approaches to encode complex, native vector geometries (e.g., polylines and polygons) into a latent space where topological errors are isolated from valid ones. A systematic performance evaluation across three study areas (Los Angeles, Munich, and Singapore) demonstrates the effectiveness and robustness of Topo4Vec, achieving a peak accuracy of 0.99 for detecting overlapping building footprints and 0.60 for overshoots and undershoots in street networks. Moreover, lessons learned from Topo4Vec shed a promising light into a scalable and autonomous GeoAI approach for large-scale vector data consistency and quality monitoring within the fast-growing geospatial data ecosystems. The code and data used in the paper are made openly available in https://figshare.com/s/612148eeb4bccadbd715.

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 / 0 minor

Summary. The manuscript proposes Topo4Vec, a GeoAI framework for automated quality assessment of geospatial vector data. It relaxes manual annotation by simulating topological errors (overlapping polygons; overshoots/undershoots in street networks), applies spatial representation learning (SRL) to embed native vector geometries into a latent space that isolates errors, and reports a systematic evaluation across three study areas (Los Angeles, Munich, Singapore) with peak accuracy 0.99 for building-footprint overlaps and 0.60 for network errors. Code and data are released openly.

Significance. If the simulation procedure produces error distributions statistically indistinguishable from real-world errors, the framework would supply a scalable, annotation-light method for vector-data consistency checking that extends beyond rule-based GIS tools and could support large-scale monitoring in diverse urban morphologies. The open release of code and data is a clear strength for reproducibility.

major comments (2)
  1. [Abstract] Abstract: the central claim that simulation enables effective SRL-based separation (and thus the reported 0.99 / 0.60 accuracies) is load-bearing, yet no quantitative comparison of geometric/topological feature distributions between simulated and authentic errors in the three study areas is supplied. Without such validation the generalization claim cannot be assessed.
  2. [Abstract] Abstract: the performance numbers are stated without reference to model architecture, SRL variant, baseline methods, statistical tests, or simulation-parameter settings, preventing evaluation of whether the cross-area robustness result is sound.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and constructive feedback. We address the two major comments point by point below. Where the comments identify gaps in the current manuscript, we propose targeted revisions to strengthen the presentation of the simulation validation and result reporting.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central claim that simulation enables effective SRL-based separation (and thus the reported 0.99 / 0.60 accuracies) is load-bearing, yet no quantitative comparison of geometric/topological feature distributions between simulated and authentic errors in the three study areas is supplied. Without such validation the generalization claim cannot be assessed.

    Authors: We agree that a quantitative comparison of geometric and topological feature distributions between simulated and real-world errors would provide stronger support for the simulation procedure. The manuscript currently justifies the simulation by describing how it replicates documented topological error types (overlaps, overshoots, undershoots) observed in practice, and demonstrates consistent performance across three morphologically distinct cities. However, no direct statistical comparison of feature distributions (e.g., overlap ratios, intersection angles, or connectivity metrics) is included. We will add this analysis in the revised manuscript, using a sample of manually verified real errors from each study area. revision: yes

  2. Referee: [Abstract] Abstract: the performance numbers are stated without reference to model architecture, SRL variant, baseline methods, statistical tests, or simulation-parameter settings, preventing evaluation of whether the cross-area robustness result is sound.

    Authors: The abstract is intentionally concise. Full details on the SRL architectures (including the specific representation learning models), variants evaluated, baseline comparisons against rule-based GIS methods, statistical tests, and simulation parameter settings are provided in the Methods and Results sections. To address the concern about the abstract, we will expand it to include brief references to the primary SRL approach, the use of cross-validation, and the key simulation parameters while remaining within length limits. revision: partial

Circularity Check

0 steps flagged

No significant circularity; empirical pipeline is self-contained

full rationale

The paper generates training examples via topological error simulation (overlaps, overshoots/undershoots), applies SRL to embed geometries, and reports accuracies on three independent geographic areas. No equations, parameters, or claims reduce by construction to the simulation rules or to self-citations; the reported metrics (0.99, 0.60) are presented as outcomes of applying the learned model to external data rather than tautological re-statements of the input generation process. The derivation chain therefore contains no self-definitional, fitted-input-renamed-as-prediction, or load-bearing self-citation steps.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 1 invented entities

The central claim depends on the domain assumption that simulated errors are representative enough for training and that SRL can separate errors in latent space; limited independent evidence is available from the abstract alone.

free parameters (1)
  • SRL model hyperparameters and simulation parameters
    Standard in ML training but unspecified in abstract; required for the reported performance.
axioms (1)
  • domain assumption Simulated topological errors sufficiently represent real-world errors to enable effective training without manual labels.
    Directly invoked to relax the manual annotation process.
invented entities (1)
  • Topo4Vec framework no independent evidence
    purpose: Automated scalable quality assessment of vector data via SRL
    New named framework introduced to combine simulation and representation learning.

pith-pipeline@v0.9.1-grok · 5817 in / 1288 out tokens · 41618 ms · 2026-06-30T09:52:49.752428+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

65 extracted references · 2 canonical work pages

  1. [1]

    Goodchild, M. F. , title =. GeoJournal , year =

  2. [2]

    and Mobasheri, A

    Senaratne, H. and Mobasheri, A. and Ali, A. L. and Capineri, C. and Haklay, M. , title =. International Journal of Geographical Information Science , year =

  3. [3]

    Rectifier nonlinearities improve neural network acoustic models , author=. Proc. icml , volume=. 2013 , organization=

    2013

  4. [4]

    , title =

    Haklay, M. , title =. Environment and planning B: Planning and design , year =

  5. [5]

    and Neis, P

    Barron, C. and Neis, P. and Zipf, A. , title =. Transactions in GIS , year =

  6. [6]

    Egenhofer, M. J. and Franzosa, R. D. , title =. International Journal of Geographical Information Systems , year =

  7. [7]

    and Di Felice, P

    Clementini, E. and Di Felice, P. and Van Oosterom, P. , title =. Advances in spatial databases , year =

  8. [8]

    and Zipf, A

    Fan, H. and Zipf, A. and Fu, Q. and Neis, P. , title =. International Journal of Geographical Information Science , year =

  9. [9]

    Girres, J. F. and Touya, G. , title =. Transactions in GIS , year =

  10. [10]

    and Pan, S

    Wu, Z. and Pan, S. and Chen, F. and Long, G. and Zhang, C. and Philip, S. Y. , title =. IEEE Transactions on Neural Networks and Learning Systems , year =

  11. [11]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Geo2vec: Shape-and distance-aware neural representation of geospatial entities , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

  12. [12]

    and Persello, C

    Tuia, D. and Persello, C. and Bruzzone, L. , title =. IEEE Geoscience and Remote Sensing Magazine , year =

  13. [13]

    Proceedings of the 31st ACM International Conference on Advances in Geographic Information Systems , pages=

    Rethink geographical generalizability with unsupervised self-attention model ensemble: A case study of openstreetmap missing building detection in africa , author=. Proceedings of the 31st ACM International Conference on Advances in Geographic Information Systems , pages=

  14. [14]

    Proceedings of the National Academy of Sciences , volume=

    Replication across space and time must be weak in the social and environmental sciences , author=. Proceedings of the National Academy of Sciences , volume=. 2021 , publisher=

    2021

  15. [15]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    Deepsdf: Learning continuous signed distance functions for shape representation , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  16. [16]

    International Journal of Geographical Information Science , volume=

    A Voronoi-based 9-intersection model for spatial relations , author=. International Journal of Geographical Information Science , volume=. 2001 , publisher=

    2001

  17. [17]

    Computers, Environment and Urban Systems , volume=

    Global building morphology indicators , author=. Computers, Environment and Urban Systems , volume=. 2022 , publisher=

    2022

  18. [18]

    International Journal of Applied Earth Observation and Geoinformation , volume=

    A review of spatially-explicit GeoAI applications in Urban Geography , author=. International Journal of Applied Earth Observation and Geoinformation , volume=. 2022 , publisher=

    2022

  19. [19]

    Remote Sensing of Environment , volume=

    Cross-city matters: A multimodal remote sensing benchmark dataset for cross-city semantic segmentation using high-resolution domain adaptation networks , author=. Remote Sensing of Environment , volume=. 2023 , publisher=

    2023

  20. [20]

    IEEE Transactions on Geoscience and Remote Sensing , volume=

    Universal domain adaptation for remote sensing image scene classification , author=. IEEE Transactions on Geoscience and Remote Sensing , volume=. 2023 , publisher=

    2023

  21. [21]

    ISPRS journal of photogrammetry and remote sensing , volume=

    A graph convolutional neural network for classification of building patterns using spatial vector data , author=. ISPRS journal of photogrammetry and remote sensing , volume=. 2019 , publisher=

    2019

  22. [22]

    International Journal of Geographical Information Science , pages=

    A graph neural network for small-area estimation: integrating spatial regularisation, heterogeneous spatial units, and Bayesian inference , author=. International Journal of Geographical Information Science , pages=. 2025 , publisher=

    2025

  23. [23]

    Nature Communications , volume=

    A spatio-temporal analysis investigating completeness and inequalities of global urban building data in OpenStreetMap , author=. Nature Communications , volume=. 2023 , publisher=

    2023

  24. [24]

    Transactions in GIS , volume=

    Improving OpenStreetMap missing building detection using few-shot transfer learning in sub-Saharan Africa , author=. Transactions in GIS , volume=. 2022 , publisher=

    2022

  25. [25]

    ISPRS Journal of Photogrammetry and Remote Sensing , volume=

    Exploration of OpenStreetMap missing built-up areas using twitter hierarchical clustering and deep learning in Mozambique , author=. ISPRS Journal of Photogrammetry and Remote Sensing , volume=. 2020 , publisher=

    2020

  26. [26]

    ISPRS Journal of Photogrammetry and Remote Sensing , volume=

    Cross-view geolocalization and disaster mapping with street-view and VHR satellite imagery: A case study of Hurricane IAN , author=. ISPRS Journal of Photogrammetry and Remote Sensing , volume=. 2025 , publisher=

    2025

  27. [27]

    AUTOCARTO-CONFERENCE- , pages=

    A critical comparison of the 4-intersection and 9-intersection models for spatial relations: formal analysis , author=. AUTOCARTO-CONFERENCE- , pages=. 1993 , organization=

    1993

  28. [28]

    International Journal of Digital Earth , volume=

    Autonomous GIS: the next-generation AI-powered GIS , author=. International Journal of Digital Earth , volume=. 2023 , publisher=

    2023

  29. [29]

    Big Earth Data , volume=

    Deep learning for processing and analysis of remote sensing big data: A technical review , author=. Big Earth Data , volume=. 2022 , publisher=

    2022

  30. [30]

    Journal of Spatial Information Science , number=

    GeoAI: Where machine learning and big data converge in GIScience , author=. Journal of Spatial Information Science , number=

  31. [31]

    Annals of GIS , volume=

    Giscience in the era of artificial intelligence: A research agenda towards autonomous gis , author=. Annals of GIS , volume=. 2025 , publisher=

    2025

  32. [32]

    Proceedings of the 32nd ACM international conference on advances in geographic information systems , pages=

    SRL: Towards a general-purpose framework for spatial representation learning , author=. Proceedings of the 32nd ACM international conference on advances in geographic information systems , pages=

  33. [33]

    Proceedings of the 3rd ACM SIGSPATIAL International Workshop on Searching and Mining Large Collections of Geospatial Data , pages=

    Random Affine Transformation Feature Representation Learning for Fast Polygon Retrieval , author=. Proceedings of the 3rd ACM SIGSPATIAL International Workshop on Searching and Mining Large Collections of Geospatial Data , pages=

  34. [34]

    Proceedings of machine learning research , volume=

    Poly2vec: Polymorphic fourier-based encoding of geospatial objects for geoai applications , author=. Proceedings of machine learning research , volume=

  35. [35]

    International Journal of Geographical Information Science , volume=

    Estimating urban functional distributions with semantics preserved POI embedding , author=. International Journal of Geographical Information Science , volume=. 2022 , publisher=

    2022

  36. [36]

    IEEE Transactions on Intelligent Transportation Systems , volume=

    Dynamic spatial-temporal representation learning for traffic flow prediction , author=. IEEE Transactions on Intelligent Transportation Systems , volume=. 2020 , publisher=

    2020

  37. [37]

    International Journal of Digital Earth , volume=

    Recognizing mixed urban functions from human activities using representation learning methods , author=. International Journal of Digital Earth , volume=. 2023 , publisher=

    2023

  38. [38]

    Annals of GIS , volume=

    Representation learning for geospatial data , author=. Annals of GIS , volume=. 2025 , publisher=

    2025

  39. [39]

    Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

    Encoding crowd interaction with deep neural network for pedestrian trajectory prediction , author=. Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

  40. [40]

    Proceedings of the IEEE/CVF International Conference on Computer Vision Workshops , pages=

    Geo-aware networks for fine-grained recognition , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision Workshops , pages=

  41. [41]

    International Journal of Geographical Information Science , volume=

    A review of location encoding for GeoAI: methods and applications , author=. International Journal of Geographical Information Science , volume=. 2022 , publisher=

    2022

  42. [42]

    Annals of GIS , volume=

    A five-year milestone: reflections on advances and limitations in GeoAI research , author=. Annals of GIS , volume=. 2024 , publisher=

    2024

  43. [43]

    Dialogues in Human Geography , volume=

    The quality of big (geo) data , author=. Dialogues in Human Geography , volume=. 2013 , publisher=

    2013

  44. [44]

    ISPRS International Journal of Geo-Information , volume=

    GeoAI for large-scale image analysis and machine vision: Recent progress of artificial intelligence in geography , author=. ISPRS International Journal of Geo-Information , volume=. 2022 , publisher=

    2022

  45. [45]

    Journal of Spatial Information Science , number=

    GeoAI for Science and the Science of GeoAI , author=. Journal of Spatial Information Science , number=. 2024 , publisher=

    2024

  46. [46]

    ISPRS journal of photogrammetry and remote sensing , volume=

    Rethinking big data: A review on the data quality and usage issues , author=. ISPRS journal of photogrammetry and remote sensing , volume=. 2016 , publisher=

    2016

  47. [47]

    ISPRS journal of Photogrammetry and Remote Sensing , volume=

    Geospatial big data handling theory and methods: A review and research challenges , author=. ISPRS journal of Photogrammetry and Remote Sensing , volume=. 2016 , publisher=

    2016

  48. [48]

    International Journal of Geographical Information Science , volume=

    Assessment of error in digital vector data using fractal geometry , author=. International Journal of Geographical Information Science , volume=. 2000 , publisher=

    2000

  49. [49]

    GeoInformatica , volume=

    Automatically conflating road vector data with orthoimagery , author=. GeoInformatica , volume=. 2006 , publisher=

    2006

  50. [50]

    GeoInformatica , volume=

    Towards general-purpose representation learning of polygonal geometries , author=. GeoInformatica , volume=. 2023 , publisher=

    2023

  51. [51]

    International journal of geographical information science , volume=

    A geographic approach for combining social media and authoritative data towards identifying useful information for disaster management , author=. International journal of geographical information science , volume=. 2015 , publisher=

    2015

  52. [52]

    Proceedings of the 32nd ACM international conference on advances in geographic information systems , pages=

    T-jepa: A joint-embedding predictive architecture for trajectory similarity computation , author=. Proceedings of the 32nd ACM international conference on advances in geographic information systems , pages=

  53. [53]

    Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining , pages=

    Controltraj: Controllable trajectory generation with topology-constrained diffusion model , author=. Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining , pages=

  54. [54]

    2024 , isbn =

    Yu, Dazhou and Hu, Yuntong and Li, Yun and Zhao, Liang , title =. 2024 , isbn =. doi:10.1145/3637528.3671738 , booktitle =

    work page doi:10.1145/3637528.3671738 2024

  55. [55]

    Transactions on Machine Learning Research , issn=

    Contrastive Graph Autoencoder for Shape-based Polygon Retrieval from Large Geometry Datasets , author=. Transactions on Machine Learning Research , issn=. 2024 , url=

    2024

  56. [56]

    International Journal of Geographical Information Science , volume =

    Xiongfeng Yan and Tinghua Ai and Min Yang and Xiaohua Tong , title =. International Journal of Geographical Information Science , volume =. 2021 , publisher =. doi:10.1080/13658816.2020.1768260 , URL =

    work page doi:10.1080/13658816.2020.1768260 2021

  57. [57]

    RSS 2016 workshop: geometry and beyond-representations, physics, and scene understanding for robotics , year=

    Signed distance fields: A natural representation for both mapping and planning , author=. RSS 2016 workshop: geometry and beyond-representations, physics, and scene understanding for robotics , year=

    2016

  58. [58]

    IEEE Access , volume=

    OpenStreetMap data quality assessment via deep learning and remote sensing imagery , author=. IEEE Access , volume=. 2019 , publisher=

    2019

  59. [59]

    International Journal of Digital Earth , volume=

    Assessing OSM building completeness for almost 13,000 cities globally , author=. International Journal of Digital Earth , volume=. 2022 , publisher=

    2022

  60. [60]

    Transactions in GIS , volume=

    Towards an open source analysis toolbox for street network comparison: Indicators, tools and results of a comparison of OSM and the official A ustrian reference graph , author=. Transactions in GIS , volume=. 2014 , publisher=

    2014

  61. [61]

    Transactions in GIS , volume=

    Towards an automated comparison of OpenStreetMap with authoritative road datasets , author=. Transactions in GIS , volume=. 2017 , publisher=

    2017

  62. [62]

    Building and Environment , volume=

    Quality of crowdsourced geospatial building information: A global assessment of OpenStreetMap attributes , author=. Building and Environment , volume=. 2023 , publisher=

    2023

  63. [63]

    Transactions in GIS , volume=

    Regional variations of context-based association rules in OpenStreetMap , author=. Transactions in GIS , volume=. 2021 , publisher=

    2021

  64. [64]

    OpenStreetMap in GIScience: experiences, research, and applications , pages=

    Quality assessment of the contributed land use information from OpenStreetMap versus authoritative datasets , author=. OpenStreetMap in GIScience: experiences, research, and applications , pages=. 2015 , publisher=

    2015

  65. [65]

    Computers, Environment and Urban Systems , volume=

    Building footprint data for countries in Africa: to what extent are existing data products comparable? , author=. Computers, Environment and Urban Systems , volume=. 2024 , publisher=

    2024