Pith. sign in

REVIEW 3 major objections 4 minor 7 references

AI and Remote Sensing for Resilient and Sustainable Built Environments: A Review of Current Methods, Open Data and Future Directions

T0 review · 3 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read This review paper claims that a critical research gap exists: almost no studies apply AI models to synthetic aperture radar (SAR) data for comprehensive bridge damage assessment, and that the review is the first to compare AI models and…

desk verdict Useful reference review with a plausible but under-documented gap claim; send to peer review after requiring a reproducible search protocol and table corrections. read the letter →

arxiv 2507.01547 v1 pith:C3PDB4TK submitted 2025-07-02 cs.CY cs.AIcs.LG

Ubada El Joulani , Tatiana Kalganova , Stergios-Aristoteles Mitoulis , Sotirios Argyroudis This is my paper
classification cs.CYcs.AIcs.LG
keywords damageassessmentsyntheticapertureradarbridgedetectiondeeplearningremotesensingcriticalinfrastructurenaturaldisasterssystematicreview
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper tries to establish that the research landscape for AI-based damage assessment is lopsided: road damage detection has many models and datasets, building damage assessment has large satellite-image benchmarks, but almost no work combines synthetic aperture radar (SAR) with deep learning to assess a bridge as a whole structure. This matters because bridges are often the hardest transport assets to inspect on the ground and the most fragile links in a network. The paper reads the scarcity as an opportunity and assembles the ingredients, from SAR variants and coherence products to open datasets, needed to fill it.

What carries the argument

The engine of the argument is a structured literature review whose comparison tables turn absence into evidence. SAR, defined as a satellite-radar imaging technique that uses antenna motion to create high-resolution images, enters through variants such as MTInSAR, InSAR, and D-TomoSAR, plus coherence products that measure correlation between radar acquisitions. The tables list which AI models and datasets exist for roads, buildings, and bridges, and the rows for SAR-plus-AI bridge assessment remain empty, which is what the paper counts as the research gap.

What would settle it

A targeted search with explicit queries such as ("SAR" OR "synthetic aperture radar") AND "bridge" AND ("damage" OR "deformation") AND ("deep learning" OR "CNN") across the same databases the paper used, covering a stated date range, would settle the claim; if it returns more than a handful of peer-reviewed studies applying AI models to SAR for whole-bridge damage assessment, the paper's central gap would be falsified.

Watch

Extended reading notes

Core claim

The paper's central claim is that a critical research gap exists: a scarcity of studies applying AI models to SAR data for comprehensive bridge damage assessment, and that a comprehensive review comparing current AI models and datasets for road and bridge damage is missing from the literature. The authors support this by systematically comparing tables of AI approaches, datasets, collection technologies, and bridge damage types, finding that bridge-related datasets focus on localised concrete defects rather than whole-structure assessment, and that existing SAR-based bridge monitoring studies using MTInSAR, InSAR, and D-TomoSAR do not yet use AI. The paper concludes with future directions: comparative model studies, expanded datasets, multi-sensor integration, AI-accelerated SAR processing, and continuous AI-driven bridge monitoring.

Load-bearing premise

The claim of scarcity rests on the completeness of Scopus and Google Scholar searches, whose exact queries and dates are not reported, so if relevant AI-plus-SAR bridge studies were missed, the gap would be an artifact of the search rather than a property of the field.

Editorial extensions

If this is right

  • Researchers seeking to assess bridge damage after disasters can treat AI applied to SAR coherence and interferometric products as an open design space.
  • A benchmark dataset comparable to xBD for buildings, but built from pre- and post-event SAR imagery of bridges, would directly address the identified gap.
  • Because optical satellite imagery is weather-dependent, SAR-based AI bridge assessment could extend damage mapping to hurricanes, floods, and conflict zones where clouds or restricted access block ground data.
  • The review's future directions imply that multi-sensor integration, combining SAR, optical imagery, and ground sensors, is needed for complete bridge health assessment.
  • Near-real-time monitoring of critical bridges will require AI to reduce the heavy processing burden of SAR data.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the gap is real, the first group to publish an open, labelled SAR bridge-damage benchmark could set the de facto standard, just as RDD2022 did for roads.
  • The scarcity may partly reflect a structural barrier rather than a lack of interest: SAR data are difficult to interpret and process, so the gap could persist until AI tools make SAR products more accessible.
  • A testable extension would be to apply existing coherence-change-detection pipelines, which already work for buildings and floods, to bridge portfolios; if those pipelines fail on bridges, radar shadow and bridge geometry, not missing effort, may be the true bottleneck.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. This manuscript is presented as a systematic literature review of AI models and datasets for damage assessment of roads, bridges, and buildings, with particular attention to the use of Synthetic Aperture Radar (SAR) data. It organizes the literature into a series of tables covering damage types, AI models, datasets, data-collection technologies, bridge damage detection methods, SAR applications, natural-hazard stressors, and damage quantification methods. The paper's central claim, stated in the Abstract and Section 1, is that there is a critical research gap: a scarcity of studies applying AI models to SAR data for comprehensive bridge damage assessment. The review concludes with recommendations for comparative model studies, expanded datasets, multi-sensor integration, and AI-assisted SAR processing.

Significance. If the gap claim is correct, the review identifies an actionable research direction and provides a useful inventory of existing datasets and models for infrastructure damage assessment. The paper's strength is its breadth of tabular synthesis, including recent works (e.g., Kopiika et al. 2025, BRIGHT, RDD2022) and its explicit discussion of open data sources. However, the central claim is an absence claim, and the evidence for absence is the collection of tables in Section 3. Because the search methodology is under-specified and some table entries are internally inconsistent, the gap claim is not currently reproducible. The paper would be substantially strengthened by a documented, rerunnable search protocol and by correction of the identified data errors.

major comments (3)
  1. [Section 2 (Methodology)] The reported search protocol is not reproducible. The manuscript states that Scopus and Google Scholar were used and that articles up to 10 years old were considered, but it gives no query strings, database-specific date ranges, numbers of records screened, or inclusion/exclusion counts. The sentence 'The search terms are the titles identified for each table corresponding to a specific research question' describes the output rather than the protocol. Because the paper's central claim is a scarcity of AI+SAR+bridge studies, the evidence is the absence of entries in the assembled tables; without a rerunnable protocol, the gap claim cannot be distinguished from a search artifact. Please provide the exact queries, search dates, record counts, and screening criteria for each table.
  2. [Tables 3, 4, and 6] Several tabular entries are internally inconsistent, which weakens the dataset inventory that supports the scarcity discussion. SDNET2018 is listed as 56,000 images in Table 4 (Roads) but as 230 images in Table 6 (Bridges), and Ni et al. (2023) appears both under Roads and under Bridges with different data sources listed. These inconsistencies affect the reliability of the comparison in Section 3.3. Please verify each table entry against the source and clarify what each entry is meant to represent (e.g., whole-bridge damage versus bridge-deck pavement damage).
  3. [Abstract and Section 1] The key qualifiers of the central claim are not operationalized. The paper asserts a scarcity of studies 'applying AI models to SAR data for comprehensive bridge damage assessment,' but it never defines what counts as 'comprehensive' as opposed to localized defect detection, nor what counts as a bridge-damage study. Table 11 already lists SAR+AI studies such as Kopiika et al. (2025), C. Wang et al. (2024), and Y. Yang et al. (2024), so the entire gap claim rests on these definitions. Please specify classification criteria and apply them consistently when screening studies.
minor comments (4)
  1. [Section 2] The phrase 'except for same cases where articles were scarce' appears to be a typo for 'except for some cases where articles were scarce.'
  2. [Section 3.2] The text states that YOLO models 'are two stage detectors (Redmon et al., 2016)'; YOLO is a single-stage detector, and the cited paper describes it as such. Please correct this description.
  3. [Section 3.7 and Table 13] 'UNOST' appears to be a typo for 'UNOSAT'; the abbreviation is spelled correctly elsewhere in the manuscript.
  4. [Table 6] The three datasets DOTA, Bridge Dataset, and AID are credited only to the IEEE GRSS platform rather than to the original dataset publications; please cite the original sources so readers can locate and assess them.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the review's gap claim is an empirical literature observation, not a derivation from its own inputs; self-citations are incidental.

full rationale

This is a systematic review, not a derivation chain: no model is fitted, no quantity is predicted, and no mathematical identity links inputs to outputs. The central claim—a scarcity of studies applying AI models to SAR data for comprehensive bridge damage assessment—is an empirical observation about the literature, supported by the tables assembled from Scopus and Google Scholar searches. The gap is not logically entailed by the inclusion criteria; it is a negative finding from the reviewed set. The search protocol is underreported (no query strings, dates, or screening counts), which is a reproducibility and correctness concern, but it is not a circularity concern under the required standard: one cannot exhibit a specific equation or definitional identity that makes the conclusion equivalent to its inputs. Self-citations (e.g., Kopiika et al. 2025, Markogiannaki et al. 2022) appear as examples of SAR applications and as background context, not as load-bearing proof of the gap, so they do not constitute circular support. The paper is self-contained as a review; no fitted parameter is renamed as a prediction, and no uniqueness theorem is imported from the authors' prior work. Therefore no significant circularity is present, and the score is 0.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

The review relies on the completeness and accuracy of its literature search and table categorization. These are assumptions about the search process rather than derived results, and at least one categorization is demonstrably incorrect.

assumptions (3)
  • domain assumption The Google Scholar and Scopus searches, with search terms not fully specified, captured a representative and sufficiently complete set of the relevant literature.
    Section 2 states the search used 'the titles identified for each table' but reports no strings, dates, or screening counts, so the review's completeness assumption is unverified.
  • ad hoc to paper Absence of a topic from the reviewed tables implies scarcity in the literature.
    The central gap conclusion follows from the absence of AI+SAR+bridge papers in Table 11, which assumes the set of reviewed papers was exhaustive rather than merely a sample.
  • domain assumption The categorization of each cited paper into infrastructure type and dataset (e.g., Ni et al. 2023 as bridges) is accurate.
    Table 3 groups Ni et al. with RDD2022, a road dataset, so the assumption is not fully satisfied.

how reviews work

0 comments
Cite this review

Pith. "Pith review of AI and Remote Sensing for Resilient and Sustainable Built Environments: A Review of Current Methods, Open Data and Future Directions." pith.science (2026). https://pith.science/paper/C3PDB4TK

@misc{pith2026250701547,
  author       = {Pith},
  title        = {Pith review of: AI and Remote Sensing for Resilient and Sustainable Built Environments: A Review of Current Methods, Open Data and Future Directions},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/C3PDB4TK}},
  note         = {Machine review of arXiv:2507.01547}
}
read the original abstract

Critical infrastructure, such as transport networks, underpins economic growth by enabling mobility and trade. However, ageing assets, climate change impacts (e.g., extreme weather, rising sea levels), and hybrid threats ranging from natural disasters to cyber attacks and conflicts pose growing risks to their resilience and functionality. This review paper explores how emerging digital technologies, specifically Artificial Intelligence (AI), can enhance damage assessment and monitoring of transport infrastructure. A systematic literature review examines existing AI models and datasets for assessing damage in roads, bridges, and other critical infrastructure impacted by natural disasters. Special focus is given to the unique challenges and opportunities associated with bridge damage detection due to their structural complexity and critical role in connectivity. The integration of SAR (Synthetic Aperture Radar) data with AI models is also discussed, with the review revealing a critical research gap: a scarcity of studies applying AI models to SAR data for comprehensive bridge damage assessment. Therefore, this review aims to identify the research gaps and provide foundations for AI-driven solutions for assessing and monitoring critical transport infrastructures.

Figures

Figures reproduced from arXiv: 2507.01547 by the authors.

Figure 13
Figure 13. FIG13.JPG [PITH_FULL_IMAGE:figures/full_fig_p031_13.png] view at source ↗
Figure 20
Figure 20. FIG20.JPG [PITH_FULL_IMAGE:figures/full_fig_p031_20.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

7 extracted references · 5 canonical work pages

  1. [802]

    O., & Hahn, M

    https://doi.org/10.1109/ICIP .2014.7025160 Ouma, Y . O., & Hahn, M. (2017). Pothole detection on asphalt pavements from 2D-colour pothole images using fuzzy c-means clustering and morphological reconstruction. Automation in Construction, 83, 196–211. https://doi.org/10.1016/J.AUTCON.2017.08.017 Pan, B., Tian, L., & Song, X. (2016). Real-time, non-contact ...

    arXiv 2017

  2. [835]

    https://doi.org/10.1109/AIEA62095.2024.10692408 Jiang, C., Zhou, Q., Lei, J., & Wang, X. (2022). A Two-Stage Structural Damage Detection Method Based on 1D-CNN and SVM. Applied Sciences (Switzerland), 12(20). https://doi.org/10.3390/app122010394 Jung, J., Kim, D. J., Vadivel, S. K. P ., & Yun, S. H. (2019). Long-Term Deflection Monitoring for Bridges Usin...

    work page arXiv 2022

  3. [956]

    (2018, May 20)

    https://doi.org/10.1109/RADAR.2017.7944341 Mouzannar, H., Rizk, Y ., & Awad, M. (2018, May 20). Damage Identification in Social Media Posts using Multimodal Deep Learning. The 15th International Conference on Information Systems for Crisis Response and Management (ISCRAM). https://idl.iscram.org/files/husseinmouzannar/2018/2129_HusseinMouzannar_etal2018.p...

    work page arXiv 2017

  4. [2054]

    https://doi.org/10.1109/TGRS.2006.872910 Radanliev, P . (2025). AI Ethics: Integrating Transparency, Fairness, and Privacy in AI Development. Applied Artificial Intelligence, 39(1), 2463722. https://doi.org/10.1080/08839514.2025.2463722 Rahman, M. R., & Thakur, P . K. (2018). Detecting, mapping and analysing of flood water propagation using synthetic aper...

    work page arXiv 2025

  5. [2867]

    Z., Lu, W., Li, Z., Khaitan, P ., & Zaytseva, V

    https://doi.org/10.3390/APP9142867 Xu, J. Z., Lu, W., Li, Z., Khaitan, P ., & Zaytseva, V . (2019). Building Damage Detection in Satellite Imagery Using Convolutional Neural Networks. ArXiv.Org. Yang, F ., Zhang, L., Yu, S., Prokhorov, D., Mei, X., & Ling, H. (2020). Feature Pyramid and Hierarchical Boosting Network for Pavement Crack Detection. IEEE Tran...

    work page arXiv 2019

  6. [3852]

    https://doi.org/10.1007/S42107-023-00748-5/TABLES/7 Abubakr, M., Rady, M., Badran, K., & Mahfouz, S. Y . (2024). Application of deep learning in damage classification of reinforced concrete bridges. Ain Shams Engineering Journal, 15(1), 102297. https://doi.org/10.1016/J.ASEJ.2023.102297 Agbaje, T. H., Abomaye-Nimenibo, N., Ezeh, C. J., Bello, A., & Olorun...

    work page arXiv 2024

  7. [5138]

    F ., Arif, Nayem, A

    https://doi.org/10.3390/S23115138 Niloy, F . F ., Arif, Nayem, A. B. S., Sarker, A., Paul, O., Amin, M. A., Ali, A. A., Zaber, M. I., & Rahman, A. M. (2021). A Novel Disaster Image Dataset and Characteristics Analysis using Attention Model. https://doi.org/10.1109/ICPR48806.2021.9412504 Oliveira, H., & Correia, P . L. (2014). CrackIT — An image processing...

    arXiv 2021

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.