Pith. sign in

REVIEW 3 major objections 4 minor 58 references

CrackUDA: Incremental Unsupervised Domain Adaptation for Improved Crack Segmentation in Civil Structures

T0 review · 3 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read CrackUDA lifts crack segmentation on unlabeled building imagery by adapting an encoder-decoder with adversarial alignment, reporting 79.83 source and 63.43 target mIoU.

desk verdict The adaptation machinery is plausible and the new dataset is useful, but the headline target gain is not a valid UDA number because target labels were used for checkpoint selection. read the letter →

arxiv 2412.15637 v1 pith:GTTF4I2X submitted 2024-12-20 cs.CV

Kushagra Srivastava , Damodar Datta Kancharla , Rizvi Tahereen , Pradeep Kumar Ramancharla , Ravi Kiran Sarvadevabhatla , Harikumar Kandath This is my paper · ORCID
classification cs.CV
keywords cracksegmentationunsuperviseddomainadaptationincrementallearningadversarialtrainingdomain-invariantfeaturesstructuralhealthmonitoringBuilddatasetencoder-decodernetwork
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 proposes CrackUDA, a two-step unsupervised domain adaptation method for crack segmentation in civil structures. The authors claim that splitting the network into domain-invariant and domain-specific parameters, adding a second decoder, and aligning features adversarially lets a model trained on labeled source images adapt to an unlabeled target domain while largely retaining source accuracy. They report 79.83 mIoU on the CrackSeg9K validation set and 63.43 mIoU on BuildCrack, a new drone-collected building crack dataset, outperforming the best baseline FADA by 0.65 and 2.7 mIoU respectively. The paper also introduces BuildCrack and shows the incremental scheme helps on held-out CrackSeg9K sub-datasets.

What carries the argument

The central object is the encoder-decoder with a residual-adapter design: the encoder contains shared domain-invariant convolutional weights and domain-specific parallel residual adapter layers (DS-RAP) plus domain-specific batch normalization (DS-BN), while two separate decoders produce domain-specific predictions. During the second step, a KL-divergence loss between the frozen first decoder's outputs and the second decoder's outputs on source images regularizes the shared weights, and a discriminator with a gradient reversal layer forces the encoder's features to be domain-indistinguishable. This combination is what the paper credits for adapting to the target while avoiding catastrophic forgetting on the source.

What would settle it

Retrain CrackUDA with checkpoints selected solely on source validation mIoU, then evaluate once on BuildCrack and compare against FADA; if the target mIoU drops to or below FADA's 60.73, the claimed 2.7 mIoU advantage comes from label-based selection rather than the adaptation mechanism.

Watch

Extended reading notes

Core claim

The central claim is that an incremental unsupervised domain adaptation framework can improve crack segmentation on a target domain without using any target labels, provided the network architecture separates domain-invariant from domain-specific parameters. In step one, a standard encoder-decoder (ERFNet backbone) is trained on labeled source images. In step two, new domain-specific adapter parameters and a second decoder are added, the first decoder is frozen, and alternating segmentation and adversarial training align source and target feature distributions through a gradient reversal layer. The authors report that this raises target mIoU on BuildCrack to 63.43 (2.7 higher than FADA) while keeping source mIoU at 79.83 (0.65 higher than FADA). They also show that removing either the KL-divergence loss on shared parameters or the adversarial alignment degrades target performance, supporting the architecture's role.

Load-bearing premise

The unsupervised framing assumes no target labels are used anywhere in training or model selection, but the protocol in Section 5.2 saves checkpoints only when mIoU increases on both source and target domains, which uses target ground truth to pick the final model.

Editorial extensions

If this is right

  • If the approach holds, crack segmentation models can be adapted to new image domains without annotating target images, reducing the cost of structural health surveys.
  • UDA methods that previously worked on driving scenes can be repurposed for thin, low-contrast structures like cracks by separating domain-invariant and domain-specific parameters.
  • The new BuildCrack dataset provides a publicly releasable benchmark for building crack segmentation under drone-imagery domain shift.
  • The incremental design suggests a path for sequentially adapting to multiple target domains without retraining from scratch, since each new domain adds its own adapter set.
  • The reported mIoU improvements, though modest, come on top of an already strong baseline, so even a few points matter in safety-critical infrastructure inspection.

Reading between the lines

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

  • The reported target gain may not be purely unsupervised: the training protocol selects checkpoints using both source and target mIoU, which uses target labels for model selection; a strictly unsupervised version would hide target labels during selection.
  • The same encoder-decoder splitting could generalize to other thin-structure segmentation tasks, such as road markings or power-line wires, where domain shift is driven by surface texture and lighting.
  • If the checkpoint-selection leak is removed, the true unsupervised advantage might be smaller than 2.7 mIoU, though the architecture's ablations suggest the alignment losses still contribute.
  • The method's reliance on a fixed source dataset and one target at a time leaves open whether adapters can be stacked for multiple targets while preserving performance on all previous ones.
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. The paper proposes CrackUDA, a two-step incremental unsupervised domain adaptation (UDA) method for binary crack segmentation. The architecture combines an encoder with shared domain-invariant parameters and domain-specific residual adapters, two domain-specific decoders, a gradient reversal layer, and an adversarial discriminator. A KL-divergence loss between the old and new decoder outputs is used to reduce catastrophic forgetting. The authors also introduce BuildCrack, a 358-image drone-captured building crack dataset. Experiments on CrackSeg9K sub-datasets and BuildCrack report a source-domain mIoU of 79.83 and a target-domain mIoU of 63.43, compared with FADA's 79.18 and 60.73, and the paper claims improvements of 0.65 and 2.7 mIoU on source and target domains respectively.

Significance. If the reported gains were valid, the paper would offer a useful incremental-UDA recipe for crack segmentation and a new public dataset for the community. The dataset contribution and the architectural idea of combining domain-invariant encoder weights with domain-specific adapters and KL-based consolidation are potentially valuable, and the ablation study is clearly presented. However, the central unsupervised target-domain claim is undermined by the target-label-based checkpoint selection protocol described in Section 5.2. Because the headline 2.7 mIoU target improvement is obtained by selecting the checkpoint that maximizes target-domain mIoU, it is not a valid UDA result as reported. The baseline comparison is also weakened by the exclusion of five non-converged baselines. These issues are load-bearing, so the manuscript in its current form cannot support the stated conclusions.

major comments (3)
  1. [Section 5.2 and Table 3] The checkpoint selection protocol invalidates the reported unsupervised target result. Section 5.2 states that in Step 2 'model checkpoints are saved only if there is an increase in mIoU scores for both the source and target domains.' Computing target-domain mIoU during training requires access to target ground-truth labels, which violates the UDA assumption that the target domain is unlabeled. Since Section 6.1 states that all 358 BuildCrack images are used for training and validation, the reported target mIoU of 63.43 in Table 3 is selected by oracle access to target labels over the training trajectory. This makes the comparison with FADA (60.73) not apples-to-apples, and the headline '2.7 mIoU improvement on target' is not a valid unsupervised result. The authors should re-run the experiments with a protocol that does not use target labels for model selection, for example by saving checkpoints based on source validation mIoU only or by fixing the checkpoint at a predetermined epoch, and report the resulting target mIoU.
  2. [Table 3] The 'state-of-the-art' comparison is not statistically or methodologically solid. Five of the nine listed baselines (AdaptSegnet, ADVENT, IAST, DAFormer, CBST) are marked as not converging and are effectively excluded from the comparison, leaving FADA, MaxSquare, DACS, and ProDA. The claim of surpassing state-of-the-art therefore rests almost entirely on a single baseline (FADA), and the source-domain gain over FADA is only 0.65 mIoU, which is within the range of run-to-run variability that the paper does not quantify. The authors should either provide a serious convergence analysis for the failed baselines, report their best obtained results with the same checkpoint-selection rule, or substantially soften the 'surpasses SOTA' claim.
  3. [Section 6.4, Table 4] The ablation study is affected by the same target-oracle checkpoint-selection issue. If the '2 Step' and '2 Step w/o KLD' rows are produced using the Section 5.2 rule of saving checkpoints only when both source and target mIoU increase, then the reported 9.93 mIoU drop on BuildCrack when removing LKLD may be confounded with the checkpoint-selection mechanism rather than reflecting the loss's genuine contribution. The paper should clarify whether the same checkpoint rule was used in ablations and, if so, re-run the ablations under a valid UDA selection protocol.
minor comments (4)
  1. [Abstract] There is a typographical error: 'without a significantdropinaccuracy' should read 'without a significant drop in accuracy'.
  2. [Equation (4)] The formula for the GRL scaling uses the symbol lambda on both sides of the equation ('lambda = 2/(1+e^{-lambda p}) - 1'), which is ambiguous; a distinct symbol such as lambda_p for the epoch-dependent value would be clearer.
  3. [Section 6.1] The statement that all 358 BuildCrack images are used for training and validation is confusing in a UDA setting, since validation with target labels is standard for evaluation but the paper should explicitly state that target labels are used only for final evaluation and never for checkpoint selection or adaptation; as written, it is contradicted by Section 5.2.
  4. [Table 1] SDNET2018 is listed as containing 1411 images with 0% cracks, yet the text says CrackSeg9K aggregates crack datasets with consistent labeling; this deserves a brief explanation to avoid the impression that a non-crack dataset was used as a segmentation source.

Circularity Check

1 steps flagged · score 6.0 of 10

Target mIoU gain in Table 3 is selected by target-label oracle during checkpointing, so the reported 2.7 mIoU improvement is not an unsupervised prediction.

  1. self definitional [Section 5.2 (Training) and Section 6.1 (Datasets and Evaluation Metrics), Table 3]
    "For both steps, the model checkpoints were saved during training. For Step 2, The model checkpoints are saved only if there is an increase in mIoU scores for both the source and target domains. ... We use all 358 BuildCrack images for training and validation."

    Step 2 is framed as UDA, where Section 3.1 defines target samples as unlabeled. Yet the checkpoint-selection rule requires computing mIoU on the target domain, which requires target ground-truth labels. Section 6.1 confirms that all 358 BuildCrack images are used for training and validation, so the labels are available during Step 2. The reported target mIoU of 63.43 is therefore the checkpoint that increased the target metric along the training trajectory, not the performance of a fixed unsupervised adaptation model. The headline gain of 2.7 mIoU over FADA (60.73) is an artifact of this oracle selection, because the baseline is not given the same target-label-based checkpoint selection.

full rationale

CrackUDA is an empirical systems paper, so there is no equation-level derivation chain to be circular: the encoder/decoder split, residual adapters, KL loss, GRL, and the BuildCrack dataset are all independently specified, and the ablations provide real component-level evidence. The source-domain gain of 0.65 mIoU is not circular because source labels are legitimately available for validation and model selection. The central target-domain claim, however, is partially circular: the Section 5.2 rule saves checkpoints only when mIoU increases on both source and target, and Section 6.1 uses all 358 BuildCrack images for training and validation. Computing target mIoU during Step 2 requires target labels, so the reported 63.43 target mIoU is a target-label-selected checkpoint maximum rather than an unsupervised prediction. Consequently, the 2.7 mIoU improvement over FADA in Table 3 is forced by the selection criterion, not by the proposed UDA method. The paper would need a fixed checkpoint-selection protocol, such as last epoch or source-validation selection, for the target improvement to be a valid unsupervised result. This is a partial circularity of the evaluation, not of the architectural derivation.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The paper posts no invented physical entities. It does rest on standard UDA assumptions, manually chosen loss weights and optimizer settings, and a violated assumption that target labels are never used. The loss weights and learning rate are the only explicit free parameters; the central quantitative claim depends more on the checkpoint-selection protocol than on any fitted constant.

free parameters (3)
  • lambda_CE = 1
    Cross-entropy weight in total segmentation loss (Eq. 10), chosen by hand; sensitivity analysis is deferred to the supplementary material.
  • lambda_KLD = 0.1
    KL-divergence weight for shared encoder regularization (Eq. 10), chosen by hand.
  • learning rate = 5e-4
    Adam learning rate used in both steps, a standard choice rather than a fitted constant.
assumptions (4)
  • domain assumption Source and target share the same binary label space (crack versus background) and aligned class semantics.
    Stated in Section 3.1; UDA transfer and the fixed decoder heads depend on this.
  • domain assumption Target samples are unlabeled and target ground truth is not used at any stage.
    Core to the UDA definition in Section 3.3; Section 5.2 uses target mIoU for checkpoint selection, so the paper internally violates this premise.
  • domain assumption Adversarial feature alignment with a gradient reversal layer reduces the domain shift for crack features.
    The GRL loss in Eq. 12 is assumed to align source and target encoder features; the paper provides only ablation evidence, not a mechanistic proof.
  • domain assumption Adding domain-specific residual adapters and a second decoder to a frozen source model preserves source performance while adapting to the target.
    This is the central architectural bet of CrackUDA, inherited from multi-domain incremental learning; its validity for crack segmentation is the empirical question being tested.

how reviews work

0 comments
Cite this review

Pith. "Pith review of CrackUDA: Incremental Unsupervised Domain Adaptation for Improved Crack Segmentation in Civil Structures." pith.science (2026). https://pith.science/paper/GTTF4I2X

@misc{pith2026241215637,
  author       = {Pith},
  title        = {Pith review of: CrackUDA: Incremental Unsupervised Domain Adaptation for Improved Crack Segmentation in Civil Structures},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GTTF4I2X}},
  note         = {Machine review of arXiv:2412.15637}
}
read the original abstract

Crack segmentation plays a crucial role in ensuring the structural integrity and seismic safety of civil structures. However, existing crack segmentation algorithms encounter challenges in maintaining accuracy with domain shifts across datasets. To address this issue, we propose a novel deep network that employs incremental training with unsupervised domain adaptation (UDA) using adversarial learning, without a significant drop in accuracy in the source domain. Our approach leverages an encoder-decoder architecture, consisting of both domain-invariant and domain-specific parameters. The encoder learns shared crack features across all domains, ensuring robustness to domain variations. Simultaneously, the decoder's domain-specific parameters capture domain-specific features unique to each domain. By combining these components, our model achieves improved crack segmentation performance. Furthermore, we introduce BuildCrack, a new crack dataset comparable to sub-datasets of the well-established CrackSeg9K dataset in terms of image count and crack percentage. We evaluate our proposed approach against state-of-the-art UDA methods using different sub-datasets of CrackSeg9K and our custom dataset. Our experimental results demonstrate a significant improvement in crack segmentation accuracy and generalization across target domains compared to other UDA methods - specifically, an improvement of 0.65 and 2.7 mIoU on source and target domains respectively.

Figures

Figures reproduced from arXiv: 2412.15637 by the authors.

Figure 1
Figure 1. BuildCrack dataset was captured by imaging building facades using a drone￾mounted camera from different angles and distances. BuildCrack has images with low contrast, occlusions, and shadows, which challenge the model’s robustness. Sample images from our building crack dataset are shown. This dataset will be made public. been explored over the years, ranging from rule-based to data-driven methods, for crack segmenta… view at source ↗
Figure 2
Figure 2. Overview of our proposed architecture (Section 5). In step 1 we train our network, M1, using the labeled source dataset S for binary segmentation. In step 2, decoder D1 and ϕs1 are frozen, and a new set of domain-specific parameters ϕs2 are added and we call this model M2. An alternating training strategy is followed in which we first train for binary segmentation on the source domain followed by adversarial trainin… view at source ↗
Figure 3
Figure 3. Qualitative results for CrackSeg9K validation set for CrackUDA and FADA [47] [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Qualitative results for BuildCrack for our network and FADA [47]. crack images collected using DJI Mavic Mini1 . All the ground-truth labels in both CrackSeg9K and BuildCrack have two class labels: background and crack. We use all 358 BuildCrack images for training and…
Figure 5
Figure 5. Figure 5: Some cases in which our approach does not perform well in CrackSeg9K and BuildCrack. 6.3 Experiments on sub-datasets of CrackSeg9K We conduct experiments on sub-datasets of CrackSeg9K, systematically exclud￾ing one sub-dataset at a time from both the training and valid…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

58 extracted references · 47 canonical work pages

  1. [1]

    Bianchi, E., Hebdon, M.: Concrete Crack Conglomerate Dataset (10 2021).https: //doi.org/10.7294/16625056.v1

    work page doi:10.7294/16625056.v1 2021

  2. [2]

    In: Computer Vision – ECCV 2018

    Chen, L.C., Zhu, Y., Papandreou, G., Schroff, F., Adam, H.: Encoder-decoder with atrous separable convolution for semantic image segmentation. In: Computer Vision – ECCV 2018. pp. 833–851. Springer International Publishing, Cham (2018)

    work page 2018

  3. [3]

    In: ICCV (October 2019)

    Chen, M., Xue, H., Cai, D.: Domain adaptation for semantic segmentation with maximum squares loss. In: ICCV (October 2019)

    work page 2019

  4. [4]

    In: Proceedings of the European Conference on Computer Vision

    Chen, X., Mottaghi, R., Liu, X., Fuchs, T., Yuille, A.: Unsupervised domain adap- tation for object detection via back-propagation. In: Proceedings of the European Conference on Computer Vision. pp. 784–800 (2018) CrackUDA 13

    work page 2018

  5. [5]

    In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition

    Chen, Y., Li, W., Sakaridis, C., Dai, D., Van Gool, L.: Domain adaptive faster r-cnn for object detection in the wild. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. pp. 3339–3348 (2018)

    work page 2018

  6. [6]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Cheng, M., Zhao, K., Guo, X., Xu, Y., Guo, J.: Joint topology-preserving and feature-refinement network for curvilinear structure segmentation. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 7147–7156 (2021)

    work page 2021

  7. [7]

    In: ICCV 2019

    Choi, J., Kim, T., Kim, C.: Self-ensembling with gan-based data augmentation for domain adaptation in semantic segmentation. In: ICCV 2019. pp. 6829–6839 (2019). https://doi.org/10.1109/ICCV.2019.00693

    work page arXiv 2019

  8. [8]

    Automation in Construction125, 103606 (2021)

    Dais, D., İhsan Engin Bal, Smyrou, E., Sarhosis, V.: Automatic crack classification and segmentation on masonry surfaces using convolutional neural networks and transfer learning. Automation in Construction125, 103606 (2021)

    work page 2021

Show all 58 references
  1. [9]

    Data in Brief21, 1664–1668 (2018)

    Dorafshan, S., Thomas, R.J., Maguire, M.: Sdnet2018: An annotated image dataset for non-contact concrete crack detection using deep convolutional neural networks. Data in Brief21, 1664–1668 (2018)

    2018

  2. [10]

    CoRR abs/2010.11929 (2020)

    Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., Uszkoreit, J., Houlsby, N.: An image is worth 16x16 words: Transformers for image recognition at scale. CoRR abs/2010.11929 (2020)

    2020 arXiv

  3. [11]

    In: IJCNN 2017

    Eisenbach, M., Stricker, R., Seichter, D., Amende, K., Debes, K., Sesselmann, M., Ebersbach, D., Stoeckert, U., Gross, H.M.: How to get pavement distress detection ready for deep learning? a systematic approach. In: IJCNN 2017. pp. 2039–2047 (2017). https://doi.org/10.1109/IJC...

    2017

  4. [12]

    In: International conference on machine learning

    Ganin, Y., Lempitsky, V.: Unsupervised domain adaptation by backpropagation. In: International conference on machine learning. pp. 1180–1189 (2015)

    2015

  5. [13]

    Ganin, Y., Ustinova, E., Ajakan, H., Germain, P., Larochelle, H., Laviolette, F., Marchand, M., Lempitsky, V.: Domain-adversarial training of neural networks (2016)

    2016

  6. [14]

    In: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision

    Garg,P.,Saluja,R.,Balasubramanian,V.N.,Arora,C.,Subramanian,A.,Jawahar, C.: Multi-domain incremental learning for semantic segmentation. In: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision. pp. 761–771 (2022)

    2022

  7. [15]

    In: 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

    Gong, R., Li, W., Chen, Y., Van Gool, L.: Dlow: Domain flow for adaptation and generalization. In: 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 2472–2481 (2019)

    2019

  8. [16]

    CoRRabs/1612.02649 (2016)

    Hoffman, J., Wang, D., Yu, F., Darrell, T.: Fcns in the wild: Pixel-level adversarial and constraint-based adaptation. CoRRabs/1612.02649 (2016)

    2016 arXiv

  9. [17]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)

    Hoyer, L., Dai, D., Van Gool, L.: DAFormer: Improving network architectures and training strategies for domain-adaptive semantic segmentation. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). pp. 9924–9935 (2022)

    2022

  10. [18]

    Applied Sciences11(13) (2021)

    Junior, G.S., Ferreira, J., Millán-Arias, C., Daniel, R., Junior, A.C., Fernandes, B.J.T.: Ceramic cracks segmentation with deep learning. Applied Sciences11(13) (2021). https://doi.org/10.3390/app11136017

    2021 doi

  11. [19]

    Khalesi, S., Ahmadi, A.: Automatic road crack detection and classification using image processing techniques, machine learning and integrated models in urban areas: A novel image binarization technique (06 2020)

    2020

  12. [20]

    Srivastava et al

    Koch, C., Georgieva, K., Kasireddy, V., Akinci, B., Fieguth, P.: A review on com- puter vision based defect detection and condition assessment of concrete and 14 K. Srivastava et al. asphalt civil infrastructure. Advanced Engineering Informatics 29(2), 196–210 (2015). https://...

    2015 doi

  13. [21]

    In: 2021 17th International Conference on Machine Vision and Applications (MVA)

    Kondo, Y., Ukita, N.: Crack segmentation for low-resolution images using joint learning with super-resolution. In: 2021 17th International Conference on Machine Vision and Applications (MVA). pp. 1–6. IEEE (2021)

    2021

  14. [22]

    IEEE Transactions on Intelligent Transportation Sys- tems 23(12), 24083–24094 (2022)

    König, J., Jenkins, M.D., Mannion, M., Barrie, P., Morison, G.: Weakly-supervised surface crack segmentation by generating pseudo-labels using localization with a classifier and thresholding. IEEE Transactions on Intelligent Transportation Sys- tems 23(12), 24083–24094 (2022)

    2022

  15. [23]

    In: Computer Vision–ECCV 2022 Workshops: Tel Aviv, Israel, October 23–27, 2022, Proceedings, Part VII

    Kulkarni, S., Singh, S., Balakrishnan, D., Sharma, S., Devunuri, S., Korlapati, S.C.R.: Crackseg9k: a collection and benchmark for crack segmentation datasets and frameworks. In: Computer Vision–ECCV 2022 Workshops: Tel Aviv, Israel, October 23–27, 2022, Proceedings, Part VII....

    2023

  16. [24]

    Neurocomputing 338, 321–348 (2019)

    Lateef, F., Ruichek, Y.: Survey on semantic segmentation using deep learning tech- niques. Neurocomputing 338, 321–348 (2019)

    2019

  17. [25]

    Ieee Access8, 114892–114899 (2020)

    Lau, S.L., Chong, E.K., Yang, X., Wang, X.: Automated pavement crack segmenta- tion using u-net-based convolutional neural network. Ieee Access8, 114892–114899 (2020)

    2020

  18. [26]

    ICML 2013 Workshop : Challenges in Representation Learning (WREPL) (07 2013)

    Lee, D.H.: Pseudo-label : The simple and efficient semi-supervised learning method for deep neural networks. ICML 2013 Workshop : Challenges in Representation Learning (WREPL) (07 2013)

    2013

  19. [27]

    IEEE Access8, 51446– 51459 (2020)

    Li, G., Wan, J., He, S., Liu, Q., Ma, B.: Semi-supervised semantic segmentation using adversarial learning for pavement crack detection. IEEE Access8, 51446– 51459 (2020). https://doi.org/10.1109/ACCESS.2020.2980086

    2020

  20. [28]

    In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition

    Liu, F., Li, X., Wang, H., Cheng, J.: Domain adaptive faster r-cnn via cross-domain marginal alignment. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. pp. 12016–12025 (2020)

    2020

  21. [29]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Liu, H., Miao, X., Mertz, C., Xu, C., Kong, H.: Crackformer: Transformer network for fine-grained crack detection. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 3783–3792 (2021)

    2021

  22. [30]

    Neurocomputing338, 139–153 (2019)

    Liu, Y., Yao, J., Lu, X., Xie, R., Li, L.: Deepcrack: A deep hierarchical feature learning architecture for crack segmentation. Neurocomputing338, 139–153 (2019)

    2019

  23. [31]

    IEEE transactions on pattern analysis and machine intelligence41(9), 1956–1970 (2018)

    Long, M., Zhu, H., Wang, J., Jordan, M.I.: Deep adaptation networks: A more general robustification scheme for deep learning. IEEE transactions on pattern analysis and machine intelligence41(9), 1956–1970 (2018)

    2018

  24. [32]

    Psychology of Learning and Motivation, vol

    McCloskey, M., Cohen, N.J.: Catastrophic interference in connectionist networks: The sequential learning problem. Psychology of Learning and Motivation, vol. 24, pp. 109–165. Academic Press (1989)

    1989

  25. [33]

    Mei, K., Zhu, C., Zou, J., Zhang, S.: Instance adaptive self-training for unsuper- vised domain adaptation (2020)

    2020

  26. [34]

    In: 2017 25th European Signal Processing Conference (EUSIPCO)

    Oliveira, H., Correia, P.L.: Road surface crack detection: Improved segmentation with pixel-based refinement. In: 2017 25th European Signal Processing Conference (EUSIPCO). pp. 2026–2030. IEEE (2017)

    2017

  27. [35]

    In: IEEE Confer- ence on Computer Vision and Pattern Recoginition (CVPR) (2020)

    Pan, F., Shin, I., Rameau, F., Lee, S., Kweon, I.S.: Unsupervised intra-domain adaptation for semantic segmentation through self-supervision. In: IEEE Confer- ence on Computer Vision and Pattern Recoginition (CVPR) (2020)

    2020

  28. [36]

    In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition

    Pei, W., Wang, Y., Vigneron, V., Wu, T.: Adversarial discriminative domain adap- tation. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. pp. 7167–7176 (2018) CrackUDA 15

    2018

  29. [37]

    Ramancharla, P., Bhalkikar, A., Velani, P., Vyas, P., Prakke, B., Patnala, N., Talyan,N.:Aprimeronrapidvisualscreening(rvs)consolidatingearthquakesafety assessment efforts in india (10 2020)

    2020

  30. [38]

    IEEE Transactions on Intelligent Transportation Systems PP, 1–10 (10 2017)

    Romera, E.,Alvarez,J.M.,Bergasa,L., Arroyo, R.:Erfnet:Efficientresidualfactor- ized convnet for real-time semantic segmentation. IEEE Transactions on Intelligent Transportation Systems PP, 1–10 (10 2017)

    2017

  31. [39]

    IEEE Access 11, 54296–54336 (2023).https://doi.org/10.1109/ACCESS.2023.3277785

    Schwonberg, M., Niemeijer, J., Termöhlen, J.A., schäfer, J.P., Schmidt, N.M., Gottschalk, H., Fingscheidt, T.: Survey on unsupervised domain adaptation for semantic segmentation for visual perception in automated driving. IEEE Access 11, 54296–54336 (2023).https://doi.org/10.1...

    2023

  32. [40]

    CoRRabs/1703.01780 (2017)

    Tarvainen, A., Valpola, H.: Weight-averaged consistency targets improve semi- supervised deep learning results. CoRRabs/1703.01780 (2017)

    2017 arXiv

  33. [41]

    In: Proceedings of the IEEE/CVF winter conference on applications of computer vision

    Tranheden, W., Olsson, V., Pinto, J., Svensson, L.: Dacs: Domain adaptation via cross-domain mixed sampling. In: Proceedings of the IEEE/CVF winter conference on applications of computer vision. pp. 1379–1389 (2021)

    2021

  34. [42]

    In: 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Tsai, Y.H., Hung, W.C., Schulter, S., Sohn, K., Yang, M.H., Chandraker, M.: Learning to adapt structured output space for semantic segmentation. In: 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 7472– 7481 (2018). https://doi.org/10.1109/CVPR.2018.00780

    2018

  35. [43]

    In: Proceedings of the IEEE/CVF international conference on computer vision

    Tsai, Y.H., Sohn, K., Schulter, S., Chandraker, M.: Domain adaptation for struc- tured output via discriminative patch representations. In: Proceedings of the IEEE/CVF international conference on computer vision. pp. 1456–1465 (2019)

    2019

  36. [44]

    In: International conference on machine learning

    Tzeng, E., Hoffman, J., Zhang, N., Saenko, K., Darrell, T.: Learning transferable features with deep adaptation networks. In: International conference on machine learning. pp. 2208–2217 (2017)

    2017

  37. [45]

    In: AIP Conference Proceedings

    Volker, A., Pahlavan, L., Blacquiere, G.: Crack depth profiling using guided wave angle dependent reflectivity. In: AIP Conference Proceedings. vol. 1650, pp. 785–

  38. [46]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Vu, T.H., Jain, H., Bucher, M., Cord, M., Pérez, P.: Advent: Adversarial entropy minimization for domain adaptation in semantic segmentation. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 2517– 2526 (2019)

    2019

  39. [47]

    In: The European Conference on Computer Vision (ECCV) (August 2020)

    Wang, H., Shen, T., Zhang, W., Duan, L., Mei, T.: Classes matter: A fine-grained adversarial approach to cross-domain semantic segmentation. In: The European Conference on Computer Vision (ECCV) (August 2020)

    2020

  40. [48]

    Automation in Construction153, 104939 (2023)

    Weng, X., Huang, Y., Li, Y., Yang, H., Yu, S.: Unsupervised domain adaptation for crack detection. Automation in Construction153, 104939 (2023)

    2023

  41. [49]

    In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition

    Xu, J., Zhu, Y., Li, X., Li, C., Wang, M., Zhang, B., Lu, H.: Unsupervised domain adaptation with adversarial residual transform networks. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. pp. 1365–1374 (2019)

    2019

  42. [50]

    IEEE Transactions on Intelligent Transportation Systems21(4), 1525–1535 (2019)

    Yang, F., Zhang, L., Yu, S., Prokhorov, D., Mei, X., Ling, H.: Feature pyramid and hierarchical boosting network for pavement crack detection. IEEE Transactions on Intelligent Transportation Systems21(4), 1525–1535 (2019)

    2019

  43. [51]

    Computer-Aided Civil and Infrastructure Engineering 32(10), 805–819 (2017)

    Zhang, A., Wang, K.C., Li, B., Yang, E., Dai, X., Peng, Y., Fei, Y., Liu, Y., Li, J.Q., Chen, C.: Automated pixel-level pavement crack detection on 3d asphalt surfaces using a deep-learning network. Computer-Aided Civil and Infrastructure Engineering 32(10), 805–819 (2017)

    2017

  44. [52]

    Srivastava et al

    Zhang, P., Zhang, B., Zhang, T., Chen, D., Wang, Y., Wen, F.: Prototypical pseudo label denoising and target structure learning for domain adaptive semantic seg- 16 K. Srivastava et al. mentation. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recogn...

    2021

  45. [53]

    arXiv preprint arXiv:2101.10979 (2021)

    Zhang, P., Zhang, B., Zhang, T., Chen, D., Wang, Y., Wen, F.: Prototypical pseudo label denoising and target structure learning for domain adaptive semantic segmen- tation. arXiv preprint arXiv:2101.10979 (2021)

    2021 arXiv

  46. [54]

    In: Proceedings of the IEEE International Conference on Computer Vision

    Zhang, Y., Qiao, Y., Liu, C., Shen, W., Wang, X.: Domain adaptive faster r-cnn with co-attention networks. In: Proceedings of the IEEE International Conference on Computer Vision. pp. 3695–3704 (2019)

    2019

  47. [55]

    Pattern Recognition Letters33(3), 227–238 (2012)

    Zou, Q., Cao, Y., Li, Q., Mao, Q., Wang, S.: Cracktree: Automatic crack detection from pavement images. Pattern Recognition Letters33(3), 227–238 (2012)

    2012

  48. [56]

    In: Proceedings of the Eu- ropean Conference on Computer Vision (ECCV)

    Zou, Y., Yu, Z., Kumar, B.V., Wang, J.: Unsupervised domain adaptation for semantic segmentation via class-balanced self-training. In: Proceedings of the Eu- ropean Conference on Computer Vision (ECCV). pp. 289–305 (2018)

    2018

  49. [57]

    In: Proceedings of the IEEE/CVF international conference on computer vision

    Zou, Y., Yu, Z., Liu, X., Kumar, B., Wang, J.: Confidence regularized self-training. In: Proceedings of the IEEE/CVF international conference on computer vision. pp. 5982–5991 (2019)

    2019

  50. [791]

    American Institute of Physics (2015)

    2015

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.