A Quantum-assisted Attention U-Net for Building Segmentation over Tunis using Sentinel-1 Data

Alessandro Sebastianelli; Babak Memar; Francesco Mauro; Luigi Russo; Paolo Gamba; Silvia Liberata Ullo

REVIEW 4 major objections 4 minor 17 references

A Quantum-assisted Attention U-Net for Building Segmentation over Tunis using Sentinel-1 Data

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

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

Pith's one-line read A frozen quantum preprocessing layer lets an Attention U-Net segment buildings on Sentinel-1 radar with 93% fewer parameters and only a small accuracy drop.

desk verdict The 93% parameter reduction claim is confounded: the frozen quanvolution layer can't explain the drop, and the base U-Net architectures are unspecified. read the letter →

arxiv 2507.13852 v1 pith:EXNAELZG submitted 2025-07-18 cs.CV eess.IV

Luigi Russo , Francesco Mauro , Babak Memar , Alessandro Sebastianelli , Silvia Liberata Ullo , Paolo Gamba This is my paper · ORCID

classification cs.CVeess.IV

keywords quantummachinelearningquanvolutionbuildingsegmentationSentinel-1SARAttentionU-NeturbanremotesensingparameterreductionTunis

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

This paper tries to establish that a frozen quantum circuit layer, applied as image preprocessing, can replace most of a segmentation network's trainable capacity without a large accuracy loss. On the urban area of Tunis, the quantum-assisted Attention U-Net reaches overall accuracy between 0.9343 and 0.9384 with 2.1 million trainable parameters, compared with 0.9491 overall accuracy and 34.8 million parameters for the classic Attention U-Net. If true, quantum-assisted feature extraction becomes a practical way to reduce computational cost for large-scale building segmentation from SAR imagery.

What carries the argument

The load-bearing component is the quanvolutional operator, a quantum-circuit analogue of convolution that transforms small image patches through angle encoding, a unitary circuit, and a measurement, with frozen (untrained) parameters. Each patch is angle-encoded into an n-qubit state, the circuit is applied, and measurements produce classical feature maps; the constraint qubits ≥ kernel size² ties the number of qubits to the kernel dimension. These feature maps become the input to an Attention U-Net whose skip connections use attention gates to weigh encoder features before merging with decoder features.

What would settle it

Retrain both models with identical U-Net architecture, depth, and number of filters, toggling only the quanvolution layer; if the non-quantum model with 2.1 million parameters already reaches roughly 0.938 overall accuracy, the reported gain cannot be attributed to the quantum preprocessing.

Watch

Extended reading notes

Core claim

The paper's central claim is that a parameter-free quanvolution preprocessing layer can carry most of the feature-extraction burden in an Attention U-Net for building segmentation from Sentinel-1 synthetic aperture radar data. On a roughly 50 km² area of Tunis with VV and VH polarizations, the quantum-assisted models achieve overall accuracy between 0.9343 and 0.9384 while training only 2.1 million parameters, against 0.9491 overall accuracy and 34.8 million parameters for the classic Attention U-Net. The paper reports this as evidence that quanvolution preserves competitive accuracy while drastically reducing computational complexity.

Load-bearing premise

The central claim assumes the classic and quantum-assisted models differ only by the quanvolution preprocessing, but the paper never specifies the base U-Net configuration for either model, so the 93% parameter reduction could partly reflect a smaller network being used for the quantum-assisted runs.

Editorial extensions

If this is right

SAR building segmentation can be run with 93% fewer trainable parameters while keeping overall accuracy within roughly 1.1 points of the full model, lowering memory and training cost.
Because the quanvolution layer is frozen, the circuit-computed feature maps could be reused across multiple downstream models without retraining the quantum part.
Circuit choice among the three tested configurations shifts accuracy by only about 0.004, suggesting the benefit is not tied to a single entanglement structure.
The remaining accuracy gap points to trainable quanvolution layers, listed as future work, as a natural next step toward matching the classic model exactly.

Reading between the lines

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

The parameter comparison bundles two changes at once: the frozen quanvolution layer and whatever U-Net configuration was used; isolating the quantum layer's contribution requires matching backbone depth, filter counts, and input size across classic and quantum runs.
Because the preprocessing layer is frozen and gives similar accuracy across circuits, the approach should port cheaply to other SAR sensors and cities once a suitable circuit is chosen.
The quanvoluted feature maps could plausibly be combined with backscatter or polarimetric descriptors as a second input branch, although Sentinel-1 GRD data lacks full polarimetry.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

Colleague,

This short conference paper applies the authors' own quanvolution layer (refs [14], [15]) to Sentinel-1 building segmentation in Tunis and reports that quantum-assisted Attention U-Nets hit OA 0.9343–0.9384 with 2.1M trainable parameters versus 0.9491 with 34.8M for the classic U-Net. Striking, but it doesn't hold up as stated.

What's genuinely useful: the study extends quanvolution to SAR data, a sensible domain, and compares three circuit variants (strongly entangled, random, basic entangled) with a fixed kernel and qubit count. The attention-gate equations are written out cleanly, and the dataset choices (Sentinel-1, Open Buildings) are public and reproducible. The literal OA numbers in Table I are plausible and internally consistent.

The soft spot is the load-bearing comparison. The quanvolution layer is frozen, so it adds zero trainable parameters. If the downstream Attention U-Net had the same architecture in both models, the trainable counts would be identical. The paper never specifies the base U-Net configuration for either run, so the 34.8M-to-2.1M drop must come from using a smaller U-Net backend, not from quanvolution. The stress-test note is right: without a matched baseline, the 93% reduction tells us nothing about the quantum layer. This is not a minor omission; it's the central claim.

Secondary gaps: no error bars, no repeated runs, no significance tests, no IoU or F1, no test-set size. The ~1-point OA gap could be noise. The study is also limited to a single city and a single image, so generalizability is unknown.

That said, the paper is honest about building on prior work, and it doesn't fabricate numbers. It's mechanically incomplete rather than deceptive. For a JURSE short paper this might pass, but for a journal it needs major revision. A serious referee could push the authors to provide the matched architecture and proper evaluation.

I'd bring this to a reading group as a cautionary tale about confounded baselines. I wouldn't cite it until the comparison is fixed, but I'd still send it out for review if I were an editor—the authors deserve the chance to correct it.

Referee Report

4 major / 4 minor

Summary. This paper proposes a quanvolutional preprocessing layer applied to Sentinel-1 VV/VH imagery of Tunis, followed by an Attention U-Net for building segmentation. Using Open Buildings 2.5D Temporal polygons as ground truth, the authors compare three quantum circuit configurations (Strongly Entangled, Random, Basic Entangled) against a standard Attention U-Net. They report overall accuracies of 0.9343–0.9384 for the quantum-assisted models with 2.1 million trainable parameters versus 0.9491 and 34.8 million for the classic model. The paper concludes that quanvolution offers a favorable accuracy–parameter trade-off. The results are explicitly preliminary and the manuscript is a short conference contribution.

Significance. If the comparison were properly controlled, the claim that a frozen quanvolution layer allows a much smaller Attention U-Net to retain near-parity accuracy would be an interesting empirical data point for quantum-assisted Earth observation. The authors use public data (Sentinel-1 via Google Earth Engine and Open Buildings) and evaluate three circuit families, which is useful breadth. However, the paper currently does not provide the architecture specification needed to interpret the parameter reduction, nor any uncertainty quantification, so the central claim is not yet established. The reported OA values at least do not show catastrophic degradation, but the evidence is insufficient to attribute the savings to quanvolution.

major comments (4)

[Section IV, Table I] The 93% trainable-parameter reduction cannot be attributed to quanvolution as described. Section III(a) states that the quanvolutional operator's parameters are frozen, so it adds zero trainable parameters. If the downstream Attention U-Net were the same in both rows, the trainable parameter count would be identical; conversely, the 2.1M model must use a much smaller U-Net backend, but the manuscript never specifies the depth, filter widths, or stage configuration for either model. Please report the full architecture for both models and, ideally, include an ablation in which the same backend is trained with and without quanvolution.
[Section IV, Table I] The accuracy comparison lacks any measure of variability. Only a single OA value is reported per configuration, with no repeated runs, no standard deviation, no test-set size, and no statistical test. With OA differences of 0.0107–0.0148 between the classic and quantum models, the claim of 'comparable' accuracy is not yet supported. Please report the number of test patches, repeated-run statistics, and at least a paired significance test or confidence intervals.
[Section IV, Table I and Section I] The paper evaluates building segmentation using only overall accuracy. In a building segmentation task, OA is dominated by the background class and can mask large changes in building-level performance. Please report IoU (or F1) for the building class, which is the standard metric for this task and would make the comparison meaningful.
[Section III(a), Eq. (1)-(2)] The quanvolutional preprocessing is under-specified for reproduction. The manuscript does not state the patch size and stride used for the quanvolution operator, the number and dimensionality of the output feature maps, the exact measurement basis, or how the 9 qubits are assigned to a 3×3 kernel. Please provide a complete specification, ideally with a pseudo-code listing or a link to released code.

minor comments (4)

[Abstract and Section III(a)] There are typographical errors, including 'indic`eate' in the abstract and 'propsoed' in Section III(a), which should be corrected.
[Section II] The patching operation does not specify how many patches were generated, the fraction used for training versus testing, or whether overlapping patches from the same image were kept in one split; please clarify to avoid potential data leakage.
[Section III(a), Eq. (2)] Equation (2) writes 'kernel size²', but the table reports kernel size as 3x3; please clarify whether this expression refers to the side length squared and define the symbols explicitly.
[Figure 4] The figure caption and text use 'quanvoluted', which is non-standard; consider 'quanvolutional feature maps' or 'quanvolved feature maps' for clarity.

Circularity Check

0 steps flagged · score 1.0 of 10

No circular derivation chain: the paper is an empirical application with independent measurements; the 93% parameter-reduction comparison is confounded but not circular.

full rationale

The paper's central claim is an empirical comparison between a classic Attention U-Net and a quantum-assisted version. The quanvolution operator is taken from the authors' prior work [14], [15], but it is used as a fixed preprocessing component and evaluated on a new Sentinel-1 Tunis dataset; the reported OA values and parameter counts are measurements, not quantities derived from the method's definition. Although the authors cite their own previous quanvolution papers as support for the claim that 'Quanvolution not only maintains model accuracy but also increases computational efficiency,' that self-citation is not load-bearing in a circular sense: the present test accuracy is independently computed, and the prior results are external empirical evidence rather than a premise that forces the outcome. The one substantive weakness is the comparison of 34.8 million trainable parameters against 2.1 million: since the quanvolutional operator is explicitly described as frozen, it cannot itself reduce the trainable parameter count, so the reduction must come from a different downstream U-Net configuration that the paper never specifies. This is a validity or confound concern, not a circularity concern, because the architecture difference is not defined in terms of the claimed result and no equation is recycled as a prediction. There is also no fitted-parameter-renamed-as-prediction step, no uniqueness theorem imported from the authors, and no known result merely renamed. Accordingly, no circular step meets the evidentiary bar required by the task, and the score is set to 1 only to acknowledge the self-referential prior-work citations that appear in the method and conclusions.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The paper's central comparison rests on hand-picked quanvolution hyperparameters, an overlapping patching scheme, the unexamined use of Open Buildings vectors as SAR ground truth, and the assumption that fewer trainable parameters means higher computational efficiency. No new physical or conceptual entities are introduced.

free parameters (5)

Number of qubits = 9
Chosen for the quanvolution layer; the paper only states the constraint qubits >= kernel size squared (Eq. 2), with no ablation or justification for 9.
Number of circuit layers = 2
Circuit depth for all three quanvolution circuits, chosen by hand without ablation.
Number of filters in quanvolution layer = 9
Number of output feature maps generated by the quanvolutional layer, chosen by hand.
Kernel size = 3x3
Receptive field of the quanvolutional operator, chosen by hand.
Patch size and stride = 256x256 with stride 128
Patching configuration for training and test generation; overlapping chips may leak information between train and test, and the choice is not justified.

assumptions (4)

domain assumption Angle encoding provides a valid and informative quantum feature map for Sentinel-1 SAR patches.
Section III-a adopts angle encoding from ref [16] without evidence specific to SAR data; the usefulness of the resulting feature maps is assumed.
domain assumption Open Buildings 2.5D Temporal building polygons are accurately aligned with Sentinel-1 GRD SAR geometry and provide valid pixel-level ground truth.
Section II uses these polygons as ground truth, but no co-registration or alignment check between the SAR image and vector footprints is described.
domain assumption Frozen, untrained random quantum circuits can extract features useful for building segmentation.
Section III-a freezes the quanvolution parameters as in ref [14]; the paper relies on this prior claim rather than demonstrating it in this setting.
domain assumption Reducing trainable parameter count translates directly into increased computational efficiency.
Sections IV and V equate a 93% parameter reduction with lower computational demand, without measuring runtime or memory use or accounting for quantum circuit simulation cost.

how reviews work

0 comments

Cite this review

Pith. "Pith review of A Quantum-assisted Attention U-Net for Building Segmentation over Tunis using Sentinel-1 Data." pith.science (2026). https://pith.science/paper/EXNAELZG

@misc{pith2026250713852,
  author       = {Pith},
  title        = {Pith review of: A Quantum-assisted Attention U-Net for Building Segmentation over Tunis using Sentinel-1 Data},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EXNAELZG}},
  note         = {Machine review of arXiv:2507.13852}
}

read the original abstract

Building segmentation in urban areas is essential in fields such as urban planning, disaster response, and population mapping. Yet accurately segmenting buildings in dense urban regions presents challenges due to the large size and high resolution of satellite images. This study investigates the use of a Quanvolutional pre-processing to enhance the capability of the Attention U-Net model in the building segmentation. Specifically, this paper focuses on the urban landscape of Tunis, utilizing Sentinel-1 Synthetic Aperture Radar (SAR) imagery. In this work, Quanvolution was used to extract more informative feature maps that capture essential structural details in radar imagery, proving beneficial for accurate building segmentation. Preliminary results indicate that proposed methodology achieves comparable test accuracy to the standard Attention U-Net model while significantly reducing network parameters. This result aligns with findings from previous works, confirming that Quanvolution not only maintains model accuracy but also increases computational efficiency. These promising outcomes highlight the potential of quantum-assisted Deep Learning frameworks for large-scale building segmentation in urban environments.

Figures

Figures reproduced from arXiv: 2507.13852 by the authors.

**Figure 2.** Diagram of the proposed methodology, illustrating [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗

**Figure 3.** Workflow of the attention gate used to enhance the performance in 2D building segmentation tasks. IV. RESULTS AND DISCUSSION The results, summarized in Table I, indicate that integrating quanvolutional operators with the Attention U-Net architecture significantly reduces the number of trainable parameters while maintaining competitive performance in terms of overall accuracy (OA). The classic Attention U-Net achiev… view at source ↗

Figures from the paper (1 more)

**Figure 4.** Figure 4: Comparison of quanvoluted feature maps, prediction, [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

17 extracted references · 15 canonical work pages

[14]

Quanv4EO: Empowering Earth Observation by means of Quanvolutional Neural Networks

A. Sebastianelli, F. Mauro, G. Ciabatti, D. Spiller, B. L. Saux, P. Gamba, and S. Ullo, “Quanv4eo: Empowering earth observation by means of quanvolutional neural networks,” arXiv preprint arXiv:2407.17108 , 2024

work page Pith review arXiv 2024
[15]

Qspecklefilter: A quantum machine learning approach for sar speckle filtering,

F. Mauro, A. Sebastianelli, M. P. Del Rosso, P. Gamba, and S. L. Ullo, “Qspecklefilter: A quantum machine learning approach for sar speckle filtering,” in IGARSS 2024 - 2024 IEEE International Geoscience and Remote Sensing Symposium , 2024, pp. 450–454

work page 2024
[1]

Automatic building extraction from google earth images under complex backgrounds based on deep instance segmentation network,

Q. Wen, K. Jiang, W. Wang, Q. Liu, Q. Guo, L. Li, and P. Wang, “Automatic building extraction from google earth images under complex backgrounds based on deep instance segmentation network,” Sensors, vol. 19, no. 2, 2019. [Online]. Available: https://www.mdpi.com/1424-8220/19/2/333

work page 2019
[2]

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, 2019

work page 2019
[3]

Semantic segmentation of satellite images with different building types using deep learning methods,

B. Amirgan and A. Erener, “Semantic segmentation of satellite images with different building types using deep learning methods,” Remote Sensing Applications: Society and Environment , vol. 34, p. 101176, 2024. [Online]. Available: https://www.sciencedirect.com/ science/article/pii/S2352938524000405

work page 2024
[4]

Large-scale individual building extraction from open-source satellite imagery via super- resolution-based instance segmentation approach,

S. Chen, Y . Ogawa, C. Zhao, and Y . Sekimoto, “Large-scale individual building extraction from open-source satellite imagery via super- resolution-based instance segmentation approach,” ISPRS Journal of Photogrammetry and Remote Sensing , vol. 195, pp. 129–152, 2023. [Online]. Available: https://www.sciencedirect.com/science/article/pii/ S0924271622002933

work page 2023
[5]

Benchmark for building segmentation on up-scaled sentinel-2 imagery,

S. Illarionova, D. Shadrin, I. Shukhratov, K. Evteeva, G. Popandopulo, N. Sotiriadi, I. Oseledets, and E. Burnaev, “Benchmark for building segmentation on up-scaled sentinel-2 imagery,” Remote Sensing, vol. 15, no. 9, 2023. [Online]. Available: https://www.mdpi.com/2072-4292/15/ 9/2347

work page 2023
[6]

U-net: Convolutional networks for biomedical image segmentation,

O. Ronneberger, P. Fischer, and T. Brox, “U-net: Convolutional networks for biomedical image segmentation,” 2015. [Online]. Available: https://arxiv.org/abs/1505.04597

arXiv 2015

Show all 17 references

[7]

A u-net architecture for building segmentation through very high resolution cosmo-skymed imagery,

B. Memar, L. Russo, and S. L. Ullo, “A u-net architecture for building segmentation through very high resolution cosmo-skymed imagery,” pp. 4653–4657, 2024

2024
[8]

Fine building segmentation in high-resolution SAR images via selective pyramid dilated network,

H. Jing, X. Sun, Z. Wang, K. Chen, W. Diao, and K. Fu, “Fine building segmentation in high-resolution SAR images via selective pyramid dilated network,” IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing , vol. 14, pp. 6608–6623, 2021

2021
[9]

Fully Complex-valued Fully Convolutional Multi-feature Fusion Network (FC 2 MFN) for Building Segmentation of InSAR images,

A. Sikdar, S. Udupa, S. Sundaram, and N. Sundararajan, “Fully Complex-valued Fully Convolutional Multi-feature Fusion Network (FC 2 MFN) for Building Segmentation of InSAR images,” pp. 581–587, 2022

2022
[10]

Development of a dual-attention U-Net model for sea ice and open water classification on SAR images,

Y . Ren, X. Li, X. Yang, and H. Xu, “Development of a dual-attention U-Net model for sea ice and open water classification on SAR images,” IEEE Geoscience and Remote Sensing Letters , vol. 19, pp. 1–5, 2021

2021
[11]

DeepMAO: Deep Multi-Scale Aware Overcomplete Network for Building Segmentation in Satellite Imagery,

A. Sikdar, S. Udupa, P. Gurunath, and S. Sundaram, “DeepMAO: Deep Multi-Scale Aware Overcomplete Network for Building Segmentation in Satellite Imagery,” pp. 487–496, 2023

2023
[12]

Analyz- ing Satellite-Derived 3D Building Inventories and Quantifying Urban Growth towards Active Faults: A Case Study of Bishkek, Kyrgyzstan,

C. Watson, J. Elliott, R. Amey, and K. Abdrakhmatov, “Analyz- ing Satellite-Derived 3D Building Inventories and Quantifying Urban Growth towards Active Faults: A Case Study of Bishkek, Kyrgyzstan,” Remote Sensing , vol. 14, no. 22, p. 5790, 2022

2022
[13]

Urban land cover classification from sentinel-2 images with quantum-classical network,

F. Fan, Y . Shi, and X. X. Zhu, “Urban land cover classification from sentinel-2 images with quantum-classical network,” in 2023 Joint Urban Remote Sensing Event (JURSE) . IEEE, 2023, pp. 1–4

2023
[16]

Data encoding patterns for quantum computing,

M. Weigold, J. Barzen, F. Leymann, and M. Salm, “Data encoding patterns for quantum computing,” in Proceedings of the 27th Conference on Pattern Languages of Programs , 2020, pp. 1–11

2020
[17]

Attention u-net: Learning where to look for the pancreas,

O. Oktay, J. Schlemper, L. L. Folgoc, M. Lee, M. Heinrich, K. Misawa, K. Mori, S. McDonagh, N. Y . Hammerla, B. Kainz, B. Glocker, and D. Rueckert, “Attention u-net: Learning where to look for the pancreas,” 2018. [Online]. Available: https://arxiv.org/abs/1804.03999

2018 arXiv

Pith tools

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

[1] [14]

Quanv4EO: Empowering Earth Observation by means of Quanvolutional Neural Networks

A. Sebastianelli, F. Mauro, G. Ciabatti, D. Spiller, B. L. Saux, P. Gamba, and S. Ullo, “Quanv4eo: Empowering earth observation by means of quanvolutional neural networks,” arXiv preprint arXiv:2407.17108 , 2024

work page Pith review arXiv 2024

[2] [15]

Qspecklefilter: A quantum machine learning approach for sar speckle filtering,

F. Mauro, A. Sebastianelli, M. P. Del Rosso, P. Gamba, and S. L. Ullo, “Qspecklefilter: A quantum machine learning approach for sar speckle filtering,” in IGARSS 2024 - 2024 IEEE International Geoscience and Remote Sensing Symposium , 2024, pp. 450–454

work page 2024

[3] [1]

Automatic building extraction from google earth images under complex backgrounds based on deep instance segmentation network,

Q. Wen, K. Jiang, W. Wang, Q. Liu, Q. Guo, L. Li, and P. Wang, “Automatic building extraction from google earth images under complex backgrounds based on deep instance segmentation network,” Sensors, vol. 19, no. 2, 2019. [Online]. Available: https://www.mdpi.com/1424-8220/19/2/333

work page 2019

[4] [2]

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, 2019

work page 2019

[5] [3]

Semantic segmentation of satellite images with different building types using deep learning methods,

B. Amirgan and A. Erener, “Semantic segmentation of satellite images with different building types using deep learning methods,” Remote Sensing Applications: Society and Environment , vol. 34, p. 101176, 2024. [Online]. Available: https://www.sciencedirect.com/ science/article/pii/S2352938524000405

work page 2024

[6] [4]

Large-scale individual building extraction from open-source satellite imagery via super- resolution-based instance segmentation approach,

S. Chen, Y . Ogawa, C. Zhao, and Y . Sekimoto, “Large-scale individual building extraction from open-source satellite imagery via super- resolution-based instance segmentation approach,” ISPRS Journal of Photogrammetry and Remote Sensing , vol. 195, pp. 129–152, 2023. [Online]. Available: https://www.sciencedirect.com/science/article/pii/ S0924271622002933

work page 2023

[7] [5]

Benchmark for building segmentation on up-scaled sentinel-2 imagery,

S. Illarionova, D. Shadrin, I. Shukhratov, K. Evteeva, G. Popandopulo, N. Sotiriadi, I. Oseledets, and E. Burnaev, “Benchmark for building segmentation on up-scaled sentinel-2 imagery,” Remote Sensing, vol. 15, no. 9, 2023. [Online]. Available: https://www.mdpi.com/2072-4292/15/ 9/2347

work page 2023

[8] [6]

U-net: Convolutional networks for biomedical image segmentation,

O. Ronneberger, P. Fischer, and T. Brox, “U-net: Convolutional networks for biomedical image segmentation,” 2015. [Online]. Available: https://arxiv.org/abs/1505.04597

arXiv 2015