pith. sign in

arxiv: 2601.16737 · v3 · submitted 2026-01-21 · 💻 cs.CV

Automated Road Crack Localization for Spatially Guided Highway Maintenance

Steffen Knoblauch , Ram Kumar Muthusamy , Pedram Ghamisi , Alexander Zipf This is my paper

Pith reviewed 2026-05-16 12:27 UTC · model grok-4.3

classification 💻 cs.CV
keywords road crack detectionYOLOhighway maintenanceOpenStreetMapaerial imagerycrack densityremote sensinginfrastructure monitoring
0
0 comments X

The pith

Combining open aerial imagery with map data lets an AI model locate highway cracks accurately enough to create a useful national maintenance index.

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

This paper shows how combining freely available airborne images with map data from OpenStreetMap can train an AI model to find cracks on highways. The model reaches solid accuracy in distinguishing cracked from intact road sections. Using these detections, the authors build a crack density index for all Swiss highways that barely relates to long-term temperature swings or traffic loads. This suggests the new index offers fresh information for deciding where to spend maintenance money as roads face more stress from weather changes.

Core claim

By fine-tuning YOLOv11 on airborne imagery aligned with OpenStreetMap data, the authors achieve F1-scores of 0.84 for crack detection and 0.97 for non-crack, and use this to compute a Swiss RHCD index that correlates weakly with land surface temperature amplitudes (r=-0.05) and traffic volume (r=0.17), demonstrating its value for guiding maintenance near urban areas and intersections.

What carries the argument

The YOLOv11 object detector, fine-tuned on georeferenced airborne imagery using OpenStreetMap for highway alignment, which enables computation of the Relative Highway Crack Density index.

Load-bearing premise

The YOLOv11 model trained on limited samples accurately detects cracks across all Swiss highways and varying conditions.

What would settle it

Independent ground-truth crack measurements from a representative sample of Swiss road segments that deviate substantially from the model's reported detection performance.

Figures

Figures reproduced from arXiv: 2601.16737 by Alexander Zipf, Pedram Ghamisi, Ram Kumar Muthusamy, Steffen Knoblauch.

Figure 1
Figure 1. Figure 1: Schematic representation of the proposed framework for scalable highway crack detection, comprising i) [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overview of the data retrieval process: downloading, masking, and tiling of highway airborne imagery. [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Overview of the modeling pipeline, including data labeling, augmentation of the training set, YOLOv11 [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Visual inspection of highway crack localization results (red = positive detection; blue = negative detection). [PITH_FULL_IMAGE:figures/full_fig_p011_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Panel A shows the Swiss highway crack localization map along with major urban centers in Switzerland [PITH_FULL_IMAGE:figures/full_fig_p012_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Panel A illustrates the spatial distribution of the RHCD index, calculated at the OSM highway segment [PITH_FULL_IMAGE:figures/full_fig_p013_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Comparison of normalized RHCD, detected from airborne imagery, with normalized LT-LST-A (left) and [PITH_FULL_IMAGE:figures/full_fig_p014_7.png] view at source ↗
read the original abstract

Highway networks are crucial for economic prosperity. Climate change-induced temperature fluctuations are exacerbating stress on road pavements, resulting in elevated maintenance costs. This underscores the need for targeted and efficient maintenance strategies. This study investigates the potential of open-source data to guide highway infrastructure maintenance. The proposed framework integrates airborne imagery and OpenStreetMap (OSM) to fine-tune YOLOv11 for highway crack localization. To demonstrate the framework's real-world applicability, a Swiss Relative Highway Crack Density (RHCD) index was calculated to inform nationwide highway maintenance. The crack classification model achieved an F1-score of $0.84$ for the positive class (crack) and $0.97$ for the negative class (no crack). The Swiss RHCD index exhibited weak correlations with Long-term Land Surface Temperature Amplitudes (LT-LST-A) (Pearson's $r\ = -0.05$) and Traffic Volume (TV) (Pearson's $r\ = 0.17$), underlining the added value of this novel index for guiding maintenance over other data. Significantly high RHCD values were observed near urban centers and intersections, providing contextual validation for the predictions. These findings highlight the value of open-source data sharing to drive innovation, ultimately enabling more efficient solutions in the public sector.

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 paper proposes a framework that combines airborne imagery and OpenStreetMap data to fine-tune YOLOv11 for localizing cracks on highways. It derives a Relative Highway Crack Density (RHCD) index for the Swiss network and reports weak correlations with long-term land surface temperature amplitudes (r = -0.05) and traffic volume (r = 0.17). The model achieves F1-scores of 0.84 for the crack class and 0.97 for the no-crack class, with elevated RHCD values near urban centers and intersections.

Significance. If the crack localization generalizes reliably, the work demonstrates a scalable use of open data for infrastructure monitoring that could reduce maintenance costs by prioritizing segments with high crack density. The weak correlations support the claim that RHCD captures information not already explained by temperature or traffic, adding a potentially useful index for public-sector planning.

major comments (2)
  1. [Abstract/Methods] Abstract and methods: The F1-scores of 0.84 (crack) and 0.97 (no-crack) are presented without any information on dataset size, train/val/test splits, geographic coverage of the imagery, imaging conditions, or cross-validation. This directly affects the central claim that the model supports a nationwide RHCD index for the entire Swiss highway network.
  2. [Results] Results section on RHCD correlations: The claim that RHCD provides added value for maintenance guidance rests on the assumption that the detected cracks are accurate across the full network; however, no evaluation on out-of-distribution data (different pavements, altitudes, lighting, or crack widths) is reported, leaving the extrapolation unsupported.
minor comments (1)
  1. [Abstract] The abstract states the model was fine-tuned but does not specify the YOLOv11 variant, learning rate, or other hyperparameters used.

Simulated Author's Rebuttal

2 responses · 1 unresolved

We thank the referee for the constructive feedback, which highlights important aspects of transparency and generalizability. We have revised the manuscript to incorporate additional methodological details and to better contextualize the limitations of our claims regarding nationwide applicability.

read point-by-point responses

  1. Referee: [Abstract/Methods] Abstract and methods: The F1-scores of 0.84 (crack) and 0.97 (no-crack) are presented without any information on dataset size, train/val/test splits, geographic coverage of the imagery, imaging conditions, or cross-validation. This directly affects the central claim that the model supports a nationwide RHCD index for the entire Swiss highway network.

    Authors: We agree that these details are essential for evaluating the reported performance metrics. The revised manuscript expands the Methods section with a full description of the dataset (including total annotated images and sources), the train/validation/test split ratios, the geographic coverage of the airborne imagery across the Swiss highway network, imaging conditions such as resolution and acquisition parameters, and the cross-validation strategy employed. These additions directly support the reliability of the F1-scores and the derivation of the RHCD index. revision: yes

  2. Referee: [Results] Results section on RHCD correlations: The claim that RHCD provides added value for maintenance guidance rests on the assumption that the detected cracks are accurate across the full network; however, no evaluation on out-of-distribution data (different pavements, altitudes, lighting, or crack widths) is reported, leaving the extrapolation unsupported.

    Authors: We acknowledge that the absence of explicit out-of-distribution testing limits the strength of claims about full-network accuracy. The open datasets available for this study do not contain sufficient diversity to enable comprehensive OOD evaluation. In the revision, we have added a dedicated Limitations subsection that explicitly discusses this gap, includes qualitative observations on varied conditions where data permitted, and moderates the language around nationwide extrapolation. We also suggest directions for future work involving more diverse imagery. This constitutes a partial revision, as full OOD testing would require new data acquisition beyond the current open-data scope. revision: partial

standing simulated objections not resolved
  • Comprehensive quantitative out-of-distribution evaluation, as the open datasets used lack the required diversity in pavements, altitudes, lighting, and crack characteristics.

Circularity Check

0 steps flagged

No circularity in derivation chain

full rationale

The paper fine-tunes YOLOv11 on airborne imagery and OSM data to localize cracks, derives the RHCD index directly from those model outputs, and reports its correlations with independent external variables (LT-LST-A and traffic volume). No load-bearing step reduces by construction to a fitted parameter, self-definition, or self-citation chain. F1 scores are standard held-out performance metrics; the index is a post-hoc aggregation compared to non-fitted external data. The derivation chain is self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 1 invented entities

The framework depends on the domain assumption that the AI model generalizes and that the derived index adds unique value, with limited free parameters explicitly stated.

free parameters (1)
  • YOLOv11 fine-tuning hyperparameters
    Parameters for training the model on the specific dataset are not specified.
axioms (1)
  • domain assumption Airborne imagery combined with OSM data is sufficient for accurate crack localization
    Invoked in the framework integration for fine-tuning.
invented entities (1)
  • Relative Highway Crack Density (RHCD) index no independent evidence
    purpose: To provide a spatially guided metric for highway maintenance prioritization
    Newly proposed index derived from model predictions without external benchmarks mentioned.

pith-pipeline@v0.9.0 · 5540 in / 1354 out tokens · 53474 ms · 2026-05-16T12:27:50.220606+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

52 extracted references · 52 canonical work pages

  1. [1]

    Di Pietro, S

    R. Di Pietro, S. Raponi, M. Caprolu, S. Cresci, Critical infrastructure, in: R. Di Pietro, S. Raponi, M. Caprolu, S. Cresci (Eds.), New Dimensions of Information Warfare, Springer International Publishing, 2021, pp. 157–196. URL:https://doi.org/10.1007/ 978-3-030-60618-3_5. doi:10.1007/978-3-030-60618-3_5

    work page doi:10.1007/978-3-030-60618-3_5 2021

  2. [2]

    Amador-Jimenez, C

    L. Amador-Jimenez, C. J. W. and, Demonstrating a correlation between infrastructure and national development, International Journal of Sustainable Development & World Ecology 19 (2012) 197–202

    work page 2012

  3. [3]

    P. H. Verburg, E. C. Ellis, A. Letourneau, A global assessment of market accessibility and market influence for global environmental change studies 6 (2011) 034019

    work page 2011

  4. [4]

    Percoco, Highways, local economic structure and urban development 16 (2016) 1035–1054

    M. Percoco, Highways, local economic structure and urban development 16 (2016) 1035–1054

    work page 2016

  5. [5]

    Vijayakumar, M

    A. Vijayakumar, M. N. Mahmood, A. Gurmu, I. Kamardeen, S. Alam, Social sustainability assessment of road infrastructure: a systematic literature review 58 (2024) 1039–1069

    work page 2024

  6. [6]

    Fifer Bizjak, A

    K. Fifer Bizjak, A. Dawson, I. Hoff, L. Makkonen, J. S. Ylhäisi, A. Carrera, The impact of climate change on the european road network 167 (2014) 281–295. Publisher: ICE Publishing

    work page 2014

  7. [7]

    S. Sen, H. Li, L. Khazanovich, Effect of climate change and urban heat islands on the deteri- oration of concrete roads 16 (2022) 100736

    work page 2022

  8. [8]

    Rioja, What is the value of infrastructure maintenance? a survey, in: G

    F. Rioja, What is the value of infrastructure maintenance? a survey, in: G. K. Ingram, K. L. Brandt (Eds.), Infrastructure and Land Policies, Lincoln Institute of Land Policy, 2013, p. Chapter 13. Commentary by Waheed Uddin

    work page 2013

  9. [9]

    Borghetti, G

    F. Borghetti, G. Beretta, N. Bongiorno, M. De Padova, Road infrastructure maintenance: Operative method for interventions’ ranking 25 (2024) 101100

    work page 2024

  10. [10]

    A. V. Kumar, Pavement surface condition assessment: a-state-of-the-art research review and future perspective 9 (2024) 470. 18

    work page 2024

  11. [11]

    Zhang, H

    N. Saeed, M. Dougherty, R. G. Nyberg, P. Rebreyend, D. Jomaa, A review of intelligent methods for unpaved roads condition assessment, in: 2020 15th IEEE Conference on Industrial Electronics and Applications (ICIEA), 2020, pp. 79–84. URL:https://ieeexplore.ieee. org/document/9248317. doi:10.1109/ICIEA48937.2020.9248317, ISSN: 2158-2297

    work page doi:10.1109/iciea48937.2020.9248317 2020

  12. [12]

    Y. Liu, J. Yao, X. Lu, R. Xie, L. Li, DeepCrack: A deep hierarchical feature learning archi- tecture for crack segmentation 338 (2019) 139–153

    work page 2019

  13. [13]

    D. Arya, H. Maeda, S. K. Ghosh, D. Toshniwal, Y. Sekimoto, RDD2022: A multi-national image dataset for automatic road damage detection, 2022. URL:http://arxiv.org/abs/ 2209.08538. doi:10.48550/arXiv.2209.08538.arXiv:2209.08538 [cs]

    work page doi:10.48550/arxiv.2209.08538.arxiv:2209.08538 2022

  14. [14]

    D. Arya, H. Maeda, S. Kumar Ghosh, D. Toshniwal, H. Omata, T. Kashiyama, Y. Sekimoto, Global road damage detection: State-of-the-art solutions, in: 2020 IEEE International Con- ference on Big Data (Big Data), 2020, pp. 5533–5539. URL:https://ieeexplore.ieee.org/ document/9377790. doi:10.1109/BigData50022.2020.9377790

    work page doi:10.1109/bigdata50022.2020.9377790 2020

  15. [15]

    Autoaudit: Mining accounting and time-evolving graphs

    V. Mandal, A. R. Mussah, Y. Adu-Gyamfi, Deep learning frameworks for pavement distress classification: A comparative analysis, in: 2020 IEEE International Conference on Big Data (Big Data), 2020, pp. 5577–5583. URL:https://ieeexplore.ieee.org/document/9378047. doi:10.1109/BigData50022.2020.9378047

    work page doi:10.1109/bigdata50022.2020.9378047 2020

  16. [16]

    Zhang, Z

    Y. Zhang, Z. Zuo, X. Xu, J. Wu, J. Zhu, H. Zhang, J. Wang, Y. Tian, Road damage detection using UAV images based on multi-level attention mechanism 144 (2022) 104613

    work page 2022

  17. [17]

    Zhong, J

    J. Zhong, J. Zhu, J. Huyan, T. Ma, W. Zhang, Multi-scale feature fusion network for pixel-level pavement distress detection 141 (2022) 104436

    work page 2022

  18. [18]

    Samadzadegan, F

    F. Samadzadegan, F. Dadrass Javan, F. Ashtari Mahini, M. Gholamshahi, F. Nex, Automatic road pavement distress recognition using deep learning networks from unmanned aerial imagery 8 (2024) 244. Number: 6 Publisher: Multidisciplinary Digital Publishing Institute

    work page 2024

  19. [19]

    D. Kang, S. S. Benipal, D. L. Gopal, Y.-J. Cha, Hybrid pixel-level concrete crack segmentation and quantification across complex backgrounds using deep learning 118 (2020) 103291. 19

    work page 2020

  20. [20]

    J.Zhong, J.Huyan, W.Zhang, H.Cheng, J.Zhang, Z.Tong, X.Jiang, B.Huang, Adeepergen- erative adversarial network for grooved cement concrete pavement crack detection 119 (2023) 105808

    work page 2023

  21. [21]

    Hacıefendioğlu, H

    K. Hacıefendioğlu, H. B. Başağa, Concrete road crack detection using deep learning-based faster r-CNN method 46 (2022) 1621–1633

    work page 2022

  22. [22]

    A. M. Roy, J. Bhaduri, DenseSPH-YOLOv5: An automated damage detection model based on DenseNet and swin-transformer prediction head-enabled YOLOv5 with attention mechanism 56 (2023) 102007

    work page 2023

  23. [23]

    G. Yu, X. Zhou, An improved YOLOv5 crack detection method combined with a bottle- neck transformer 11 (2023) 2377. Number: 10 Publisher: Multidisciplinary Digital Publishing Institute

    work page 2023

  24. [24]

    F. Wan, C. Sun, H. He, G. Lei, L. Xu, T. Xiao, YOLO-LRDD: a lightweight method for road damage detection based on improved YOLOv5s 2022 (2022) 98

    work page 2022

  25. [25]

    J. Zeng, H. Zhong, YOLOv8-PD: an improved road damage detection algorithm based on YOLOv8n model 14 (2024) 12052. Publisher: Nature Publishing Group

    work page 2024

  26. [26]

    Y. Zhao, L. Zhou, X. Wang, F. Wang, G. Shi, Highway crack detection and classification using UAV remote sensing images based on CrackNet and CrackClassification 13 (2023) 7269. Number: 12 Publisher: Multidisciplinary Digital Publishing Institute

    work page 2023

  27. [27]

    M. Z. Bashar, C. Torres-Machi, Exploring the capabilities of optical satellite imagery in evaluating pavement condition (2022) 108–115. Publisher: American Society of Civil Engineers

    work page 2022

  28. [28]

    URL:https://www.swisstopo.admin.ch/de

    swisstopo, 2024. URL:https://www.swisstopo.admin.ch/de

    work page 2024

  29. [29]

    URL:https://data.geo.admin.ch/

    data.geo.admin.ch, 2024. URL:https://data.geo.admin.ch/

    work page 2024

  30. [30]

    URL:https://www.bfs.admin.ch/bfs/ en/home/statistics/mobility-transport/transport-infrastructure-vehicles/ network-length.html

    Infrastructure and network length, 2024. URL:https://www.bfs.admin.ch/bfs/ en/home/statistics/mobility-transport/transport-infrastructure-vehicles/ network-length.html

    work page 2024

  31. [31]

    URL:https://download.geofabrik.de/europe/ switzerland.html

    Geofabrik download server, 2024. URL:https://download.geofabrik.de/europe/ switzerland.html. 20

    work page 2024

  32. [32]

    URL:https://dashboard.ohsome.org/#backend=ohsomeApi

    ohsome - dashboard, 2024. URL:https://dashboard.ohsome.org/#backend=ohsomeApi

    work page 2024

  33. [33]

    S. W. I. swissinfo.ch, Record-breaking traffic jams on swiss roads, 2024. URL:https://www. swissinfo.ch/eng/swiss-politics/record-breaking-traffic-jams-on-swiss-roads/ 80936609

    work page 2024

  34. [34]

    URL:https://data-explorer.oecd.org/vis?df[ds] =DisseminateFinalDMZ&df[id]=DSD_TRENDS@DF_TRENDS&df[ag]=OECD.ITF&dq=CHE.A

    OECD data explorer, 2022. URL:https://data-explorer.oecd.org/vis?df[ds] =DisseminateFinalDMZ&df[id]=DSD_TRENDS@DF_TRENDS&df[ag]=OECD.ITF&dq=CHE.A. CONTAINER+FREIGHT+PASSENGER.PASKM.ROAD.ROAD.&lom=LASTNPERIODS&lo=5&to[TIME_ PERIOD]=false&vw=tb

    work page 2022

  35. [35]

    Road freight transported in switzerland 2008-2021 | statista,

    work page 2008

  36. [36]

    URL:https://www.statista.com/statistics/435544/ switzerland-tonne-kilometres-of-freight-transported-by-road/

    work page

  37. [37]

    URL:https://www.are.admin.ch/are/de/ home.html

    Bundesamt für raumentwicklung ARE, 2024. URL:https://www.are.admin.ch/are/de/ home.html

    work page 2024

  38. [38]

    A. M. Fischer, K. M. Strassmann, M. Croci-Maspoli, A. M. Hama, R. Knutti, S. Kotlarski, C. Schär, C. Schnadt Poberaj, N. Ban, M. Bavay, U. Beyerle, D. N. Bresch, S. Brönnimann, P. Burlando, A. Casanueva, S. Fatichi, I. Feigenwinter, E. M. Fischer, M. Hirschi, M. A. Liniger, C. Marty, I. Medhaug, N. Peleg, M. Pickl, C. C. Raible, J. Rajczak, O. Rössler, S....

    work page 2022

  39. [39]

    URL:https://www

    Slippery roads warnings in switzerland - natural hazards portal, 2025. URL:https://www. natural-hazards.ch/home/current-natural-hazards/slippery-roads.html

    work page 2025

  40. [40]

    ge- ological survey, 2024

    Landsat 8-9 collection 2 level 2 science product guide | u.s. ge- ological survey, 2024. URL:https://www.usgs.gov/media/files/ landsat-8-9-collection-2-level-2-science-product-guide

    work page 2024

  41. [41]

    Gorelick, M

    N. Gorelick, M. Hancher, M. Dixon, S. Ilyushchenko, D. Thau, R. Moore, Google earth engine: Planetary-scale geospatial analysis for everyone 202 (2017) 18–27. 21

    work page 2017

  42. [42]

    Q. Weng, D. Lu, J. Schubring, Estimation of land surface temperature–vegetation abundance relationship for urban heat island studies 89 (2004) 467–483

    work page 2004

  43. [43]

    Wolhuter, Geometric Design of Roads Handbook, CRC Press, 2015

    K. Wolhuter, Geometric Design of Roads Handbook, CRC Press, 2015. doi:10.1201/b18344

    work page doi:10.1201/b18344 2015

  44. [44]

    Casali, H

    Y. Casali, H. R. Heinimann, A topological characterization of flooding impacts on the zurich road network 14 (2019) e0220338. Publisher: Public Library of Science

    work page 2019

  45. [45]

    F. R. O. FEDRO, Facts and figures 2023/2024, 2023. URL:https://www.astra.admin.ch/ astra/en/home/dokumentation/strassen-verkehr.html

    work page 2023

  46. [46]

    URL:https://www.astra.admin.ch/astra/ de/home/das-astra/kontakt/gebietseinheiten.html

    The territorial units of the national roads, 2024. URL:https://www.astra.admin.ch/astra/ de/home/das-astra/kontakt/gebietseinheiten.html

    work page 2024

  47. [47]

    URL:https://www.efk.admin.ch/en/audit/ economic-efficiency-of-territorial-units-and-their-supervision-federal-roads-office/

    Economic efficiency of territorial units and their super- vision, 2024. URL:https://www.efk.admin.ch/en/audit/ economic-efficiency-of-territorial-units-and-their-supervision-federal-roads-office/

    work page 2024

  48. [48]

    B. f. S. ASTRA, Unterhaltsplanung nationalstrassen, 2024. URL:https://www.astra. admin.ch/astra/de/home/themen/nationalstrassen/baustellen/wissenswertes/ unterhaltsplanung-nationalstrassen.html

    work page 2024

  49. [49]

    URL:https://concrete-ly

    Operation and maintenance – swiss highways - concretely, 2024. URL:https://concrete-ly. com/en/operation-maintenance-swiss-highways/

    work page 2024

  50. [50]

    STRABAG switzerland goes all-digital with trimble on pilot swiss road reconstruction project,

    work page

  51. [51]

    URL:https://fieldtech.trimble.com/resources/construction-industry-2/ strabag-switzerland-goes-all-digital-with-trimble-on-pilot-swiss-road-reconstruction-project-2

    work page

  52. [52]

    URL:https://www.astra.admin.ch/ astra/de/home/fachleute/weitere-bereiche/fachanwendungen/kuba.html

    KUBA - engineering structures and tunnels, 2025. URL:https://www.astra.admin.ch/ astra/de/home/fachleute/weitere-bereiche/fachanwendungen/kuba.html. 22

    work page 2025