Pith. sign in

REVIEW 3 major objections 4 minor 62 references

Adversarial Learning and Self-Teaching Techniques for Domain Adaptation in Semantic Segmentation

T0 review · 3 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read Combining adversarial learning with a confidence-weighted, region-growing self-teaching loss lets segmentation networks trained on synthetic street scenes adapt to real scenes without any real labels, raising GTA5-to-Cityscapes mIoU from…

desk verdict Solid incremental UDA method with useful ablations, but the state-of-the-art claim is contradicted by stronger baselines the paper itself cites. read the letter →

arxiv 1909.00781 v2 pith:E3OKOAIV submitted 2019-09-02 cs.CV cs.LGeess.IV

Umberto Michieli , Matteo Biasetton , Gianluca Agresti , Pietro Zanuttigh This is my paper · ORCID
classification cs.CVcs.LGeess.IV
keywords domainadaptationsemanticsegmentationadversariallearningself-teachingsynthetic-to-realregiongrowingdiscriminatorconfidenceclass-frequencyweighting
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 an unsupervised domain adaptation method that lets a semantic segmentation network trained on synthetic street scenes adapt to real street scenes without any real-world labels. It combines an adversarial discriminator that scores each pixel of the predicted segmentation with a self-teaching loss in which the network is trained on its own high-confidence predictions on unlabeled real images. The self-teaching is refined by a region-growing step that expands confident pixels to neighboring pixels with similar predicted class scores, and by class-frequency weighting that upweights rare classes. On the GTA5-to-Cityscapes benchmark the method lifts mIoU from 27.9 to 33.3, with consistent gains on SYNTHIA and Mapillary. The authors argue the approach reaches state-of-the-art accuracy for this task.

What carries the argument

The central object is the fully convolutional discriminator D, which outputs a per-pixel confidence map instead of a single real/fake value. Its output on the generator's prediction for a real image is interpreted as a measure of how trustworthy that prediction is. A threshold mask selects confident pixels, then a region-growing procedure expands each confident pixel to adjacent pixels whose predicted class score exceeds a high threshold, producing a refined confidence mask $m^R_{T_u}$. The self-teaching loss multiplies the generator's one-hot prediction by this mask, by the discriminator confidence, and by a class-frequency weight $W^s_c$ computed from the source labels, so that pseudo-labels on unlabeled real data train the network with a soft, class-balanced confidence weighting.

What would settle it

Measure, on held-out target images with ground truth, whether pixels with high discriminator confidence are more often correctly segmented than low-confidence pixels; if the correlation is weak or negative, the self-teaching signal is driving the network with untrustworthy pseudo-labels.

Watch

Extended reading notes

Core claim

The central claim is that the reliability signal provided by a fully convolutional discriminator can drive effective self-teaching for domain adaptation, provided the confidence maps are expanded by region growing and reweighted by class frequency. The paper shows that training the generator with three losses—supervised cross-entropy on synthetic data, adversarial loss on both domains, and the confidence-weighted self-teaching loss on real data—raises accuracy by about 5–6 mIoU points across four source-target pairs. The method is agnostic to the segmentation architecture; experiments use DeepLab v2. The ablation attributes the improvement to the combination of the discriminator weighting and region growing, and shows the class-frequency weights stabilize performance.

Load-bearing premise

The load-bearing premise is that the discriminator's per-pixel output reliably indicates whether the segmentation network's prediction on real data is correct; if the discriminator is confidently wrong, the pseudo-labels will reinforce those errors.

Editorial extensions

If this is right

  • The framework works with any segmentation backbone, so stronger encoders can be dropped in without changing the adaptation scheme.
  • Rare and small classes such as traffic lights, poles, and signs receive disproportionately less pseudo-supervision; the class-frequency weighting directly counteracts this and generalizes to any imbalanced segmentation task.
  • Because the region-growing extension recovers edges and small objects, the approach is likely to help in other dense prediction settings where boundaries matter.
  • Consistent gains of roughly 5–6 mIoU across four synthetic-to-real pairs suggest the recipe transfers across datasets rather than overfitting one distribution.

Reading between the lines

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

  • One testable extension is to estimate the class-frequency weights from the target domain's predicted label distribution instead of the source labels, since the paper's Eq. 7 uses only source statistics; this could correct the mismatch when source and target class priors differ.
  • The soft weighting replaces the hard threshold of the prior version, but the ablation shows it alone is unstable on SYNTHIA; this suggests the stability comes from the region-growing expansion, which could be studied as a general regularizer for self-training.
  • The same discriminator-confidence weighting could be applied to other dense prediction tasks such as monocular depth estimation or instance segmentation, where pseudo-labels are equally prone to systematic error.
  • The method does not require any target labels but still requires careful tuning of three loss weights; an automated weight scheme based on validation-free uncertainty is an obvious next step.
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.

Referee Report

3 major / 4 minor

Summary. The manuscript proposes an unsupervised domain adaptation (UDA) method for semantic segmentation that combines a supervised cross-entropy loss on synthetic data with an adversarial learning module and a self-teaching loss on unlabeled real data. The self-teaching loss weights pixels by the discriminator's confidence, expands reliable regions via a region-growing step, and applies class-frequency weighting estimated from the source domain. Experiments are reported for adaptation from GTA5 and SYNTHIA to Cityscapes and Mapillary, with mIoU improvements over the supervised baseline and over several prior methods, and the paper claims state-of-the-art performance.

Significance. If the empirical claims are supported, the method would offer a practical, architecture-agnostic way to adapt synthetic-trained segmentation networks to real scenes without target labels. The ablation study is detailed and the authors provide code, which are notable strengths. The central weakness is that the state-of-the-art claim is not supported by the comparisons included: the paper omits several contemporaneous baselines that it itself cites and that report higher mIoU on the same benchmarks. The technical components are plausible and may yield genuine incremental gains, but the paper's positioning and the evidence for its headline claim need to be revised.

major comments (3)
  1. [Section V-A, Tables II and III] The paper claims state-of-the-art results in the Abstract and Section I, but the comparison set in Tables II–V includes only [48], [5], [16], and the authors' prior work [3]. The Related Work cites AdaptSegNet [50], CBST [53], and ADVENT [57], all of which are contemporaneous methods that report substantially higher mIoU on the same GTA5-to-Cityscapes and SYNTHIA-to-Cityscapes benchmarks, with the same DeepLab-v2 backbone in common cases. Because the state-of-the-art claim is the headline contribution, this omission is load-bearing. The authors should add these baselines to the comparison tables and, if the reported numbers do not surpass them, remove or qualify the state-of-the-art wording.
  2. [Section V-C, Table VII] The hyperparameters ws, wt, and w' are selected by evaluating on the Cityscapes validation set (the reported defaults differ between the GTA5 and SYNTHIA source settings), and the same Cityscapes validation set is used for the final mIoU numbers in Tables II and III. This selection procedure means the final comparison is not a fixed-protocol evaluation but a tuned one, which weakens the comparison with prior work. Please disclose this tuning procedure in the main text, and either report results with a fixed default set of weights or evaluate the chosen weights on a held-out target dataset.
  3. [Section V, Tables II–V] All experimental results appear to be based on a single run; no standard deviations, error bars, or repeated-seed evaluations are reported. Some of the differences that support the component claims are small (e.g., 33.3 vs. 33.1 when class weighting is removed on GTA5, and 31.3 vs. 30.2 on SYNTHIA). Without variance estimates, it is not possible to assess whether these differences are significant. Please report means and standard deviations over at least three runs with different random seeds, or justify why this is computationally infeasible.
minor comments (4)
  1. [Section III, Eqs. (4) and (7)] The class-frequency weighting is written as W_c^s in Eq. (4) and in the text, but Eq. (7) defines W_c^t. The superscript should be made consistent throughout.
  2. [Section V-C, Table VI] The checkmark notation in Table VI is ambiguous: several rows display the same number of checkmarks but different mIoU values, so the reader cannot tell which components are enabled in each row. Please label each row explicitly with the set of enabled components.
  3. [Section V-C] The sensitivity of the thresholds Tu and TR is not analyzed, although region growing is presented as a key novelty. A short ablation over these two thresholds would help the reader understand how robust the method is to their choice.
  4. [Section I and Section V] The phrase 'state-of-the-art' is used in the Abstract and Introduction, but the experimental section does not specify the exact training protocol (number of iterations, batch composition, etc.) used for the omitted baselines. Please state how the numbers for all compared methods were obtained or cite the original sources clearly.

Circularity Check

1 steps flagged · score 2.0 of 10

No derivation-chain circularity; one minor selection-on-validation issue in the loss-weight tuning.

  1. fitted input called prediction [Section III (Eq. 8), Section IV (Cityscapes protocol), Table VII]
    "Finally, the overall loss function for the training of G is a weighted average of the three losses, i.e.: Lf ull = LG,1 + ws,tLs,t G,2 + w′LG,3 (8) We empirically set the weighting parameters as specified in Section V. ... Different values of the balancing hyper-parameters of Eq. 8 are reported applying various scaling factors to each parameter. The default parameters are ws = 10−2, wt = 10−4, w′ = 10−3 when adapting from GTA5 ... while the 500 images in the original validation set have been used as a test set."

    The balancing weights in Eq. 8 are not derived from the model or from source-domain quantities; they are selected by scanning scaling factors on the Cityscapes validation split (Table VII) and the setting with the best mIoU is then reported as the method's result. The same 500-image validation split is used as the reported test set. Thus the headline numbers 33.3 and 31.3 are partly the outcome of a hyperparameter search over the evaluation set, i.e., a fitted input reported as an out-of-sample performance prediction. This is a real but mild circular burden: the improvement over the supervised baseline is still not forced by construction, and the similar Mapillary gains give some external support.

full rationale

The paper's central technical chain is not circular in the derivation sense. The self-teaching loss Eq. 4 explicitly uses the model's own one-hot argmax as pseudo-labels, but this is an acknowledged self-training mechanism and the final evaluation is against external Cityscapes and Mapillary ground truth, so no quantity is equivalent to its input by construction. The adversarial and class-frequency terms are also defined from source labels or discriminator outputs rather than from the reported accuracy. The only noteworthy circularity burden is the empirical selection of the loss-balancing hyperparameters on the same Cityscapes validation split that serves as the test set; Table VII reports the resulting mIoU for different scalings and the best/default configuration is presented as the method's result. That is a mild fitted-input-called-prediction effect and justifies a score of 2. The paper's broader 'state-of-the-art' claim is not supported by the baselines actually tabulated, since it cites but omits AdaptSegNet [50], CBST [53], and ADVENT [57] from the comparison, but this is an empirical-support problem rather than a circularity and does not raise the circularity score.

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

The central claim rests on standard deep-learning training assumptions plus several hand-set thresholds and loss weights. The method is empirical: there is no derivation from first principles. The most consequential free parameters are the two confidence thresholds and the three loss-balancing weights, one set of which is chosen on the Cityscapes validation set that is also used to report the main result.

free parameters (3)
  • Tu (confidence threshold) = 0.2
    Threshold in Eq. 5 for selecting initially confident pixels from the discriminator output; hand-set in Section III.
  • TR (region growing threshold) = 1 - 1e-5
    Threshold for expanding confident regions to neighboring pixels with high class probability (Section III).
  • Loss weights ws, wt, w' = GTA5: 1e-2, 1e-4, 1e-3; SYNTHIA: 1e-2, 1e-3, 1e-1
    Weights in Eq. 8 balancing supervised, adversarial, and self-teaching losses; chosen on Cityscapes validation (Section V.C).
assumptions (4)
  • ad hoc to paper The discriminator's per-pixel output is a reliable proxy for the correctness of the segmenter's prediction.
    Used to construct the self-teaching weight in Eq. 4 and region growing mask in Eq. 6; central to the adaptation mechanism.
  • domain assumption Source-domain class frequencies approximate target-domain frequencies for the class weighting W_c^s.
    Class weights in Eq. 7 are computed on synthetic ground truth; the paper assumes they are transferable and later tests robustness to mismatch.
  • domain assumption Neighboring pixels with high class probability for the same class belong to the same semantic region.
    Underlies the region-growing expansion in Section III; can fail at object boundaries or for connected objects of different classes.
  • domain assumption DeepLab v2 with ResNet-101 pretrained on MSCOCO is an appropriate segmenter for the task.
    The generator architecture is fixed; results may not transfer to other backbones.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Adversarial Learning and Self-Teaching Techniques for Domain Adaptation in Semantic Segmentation." pith.science (2026). https://pith.science/paper/E3OKOAIV

@misc{pith2026190900781,
  author       = {Pith},
  title        = {Pith review of: Adversarial Learning and Self-Teaching Techniques for Domain Adaptation in Semantic Segmentation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/E3OKOAIV}},
  note         = {Machine review of arXiv:1909.00781}
}
read the original abstract

Deep learning techniques have been widely used in autonomous driving systems for the semantic understanding of urban scenes. However, they need a huge amount of labeled data for training, which is difficult and expensive to acquire. A recently proposed workaround is to train deep networks using synthetic data, but the domain shift between real world and synthetic representations limits the performance. In this work, a novel Unsupervised Domain Adaptation (UDA) strategy is introduced to solve this issue. The proposed learning strategy is driven by three components: a standard supervised learning loss on labeled synthetic data; an adversarial learning module that exploits both labeled synthetic data and unlabeled real data; finally, a self-teaching strategy applied to unlabeled data. The last component exploits a region growing framework guided by the segmentation confidence. Furthermore, we weighted this component on the basis of the class frequencies to enhance the performance on less common classes. Experimental results prove the effectiveness of the proposed strategy in adapting a segmentation network trained on synthetic datasets, like GTA5 and SYNTHIA, to real world datasets like Cityscapes and Mapillary.

Figures

Figures reproduced from arXiv: 1909.00781 by the authors.

Figure 1
Figure 1. Architecture of the proposed framework. The optimization is guided by a discriminator loss and 3 losses for the generator: a standard cross-entropy loss on synthetic data (LG,1), an adversarial loss (L s,t G,2 ) and a self-teaching loss for unlabeled real data (LG,3). Best viewed in colors. discriminator refined by a region growing procedure that ex￾ploits pixel aggregation to improve the confidence estimation. The … view at source ↗
Figure 2
Figure 2. Semantic segmentation of some sample scenes extracted from the Cityscapes (a) and Mapillary (b) validation sets. The [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

62 extracted references · 54 canonical work pages

  1. [48]

    FCNs in the wild: Pixel-level adversarial and constraint-based adaptation,

    J. Hoffman, D. Wang, F. Yu, and T. Darrell, “FCNs in the wild: Pixel-level adversarial and constraint-based adaptation,” arXiv preprint arXiv:1612.02649, 2016

    arXiv 2016

  2. [5]

    Adversarial learning for semi-supervised semantic segmentation,

    W.-C. Hung, Y .-H. Tsai, Y .-T. Liou34, Y .-Y . Lin, and M.-H. Yang15, “Adversarial learning for semi-supervised semantic segmentation,” in Proceedings of the British Machine Vision Conference , 2018

    work page 2018

  3. [16]

    Curriculum domain adaptation for semantic segmentation of urban scenes,

    Y . Zhang, P. David, and B. Gong, “Curriculum domain adaptation for semantic segmentation of urban scenes,” in Proceedings of International Conference on Computer Vision , 2017, pp. 2020–2030

    work page 2017

  4. [3]

    Unsupervised Domain Adaptation for Semantic Segmentation of Urban Scenes,

    M. Biasetton, U. Michieli, G. Agresti, and P. Zanuttigh, “Unsupervised Domain Adaptation for Semantic Segmentation of Urban Scenes,” in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition Workshops, 2019

    work page 2019

  5. [50]

    Learning to adapt structured output space for semantic segmentation,

    Y .-H. Tsai, W.-C. Hung, S. Schulter, K. Sohn, M.-H. Yang, and M. Chandraker, “Learning to adapt structured output space for semantic segmentation,” in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, 2018, pp. 7472–7481

    work page 2018

  6. [53]

    Unsupervised domain adaptation for semantic segmentation via class-balanced self-training,

    Y . Zou, Z. Yu, B. Vijaya Kumar, and J. Wang, “Unsupervised domain adaptation for semantic segmentation via class-balanced self-training,” in Proceedings of European Conference on Computer Vision , 2018, pp. 289–305

    work page 2018

  7. [57]

    Advent: Adversarial entropy minimization for domain adaptation in semantic segmentation,

    T.-H. Vu, H. Jain, M. Bucher, M. Cord, and P. P ´erez, “Advent: Adversarial entropy minimization for domain adaptation in semantic segmentation,” in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, 2019, pp. 2517–2526

    work page 2019

  8. [1]

    Playing for data: Ground truth from computer games,

    S. R. Richter, V . Vineet, S. Roth, and V . Koltun, “Playing for data: Ground truth from computer games,” in Proceedings of European Conference on Computer Vision , B. Leibe, J. Matas, N. Sebe, and M. Welling, Eds., vol. 9906. Springer International Publishing, 2016, pp. 102–118

    work page 2016

Show all 62 references
  1. [2]

    The synthia dataset: A large collection of synthetic images for semantic segmentation of urban scenes,

    G. Ros, L. Sellart, J. Materzynska, D. Vazquez, and A. M. Lopez, “The synthia dataset: A large collection of synthetic images for semantic segmentation of urban scenes,” in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition , 2016, pp. 3234–3243

    2016

  2. [4]

    Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected crfs,

    L.-C. Chen, G. Papandreou, I. Kokkinos, K. Murphy, and A. L. Yuille, “Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolution, and fully connected crfs,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 40, pp. 834–848, 2018

    2018

  3. [6]

    Semi- supervised automatic segmentation of layer and fluid region in retinal optical coherence tomography images using adversarial learning,

    X. Liu, J. Cao, T. Fu, Z. Pan, W. Hu, K. Zhang, and J. Liu, “Semi- supervised automatic segmentation of layer and fluid region in retinal optical coherence tomography images using adversarial learning,” IEEE Access, vol. 7, pp. 3046–3061, 2019

    2019

  4. [7]

    A survey on deep learn- ing techniques for image and video semantic segmentation,

    A. Garcia-Garcia, S. Orts-Escolano, S. Oprea, V . Villena-Martinez, P. Martinez-Gonzalez, and J. Garcia-Rodriguez, “A survey on deep learn- ing techniques for image and video semantic segmentation,”Applied Soft Computing, vol. 70, pp. 41 – 65, 2018

    2018

  5. [8]

    Fully convolutional networks for semantic segmentation,

    J. Long, E. Shelhamer, and T. Darrell, “Fully convolutional networks for semantic segmentation,” in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition , 2015, pp. 3431–3440

    2015

  6. [9]

    Multi-scale context aggregation by dilated convolutions,

    F. Yu and V . Koltun, “Multi-scale context aggregation by dilated convolutions,” inInternational Conference on Learning Representations, 2016

    2016

  7. [10]

    Pyramid scene parsing network,

    H. Zhao, J. Shi, X. Qi, X. Wang, and J. Jia, “Pyramid scene parsing network,” in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, 2017, pp. 2881–2890

    2017

  8. [11]

    Crdoco: Pixel- level domain transfer with cross-domain consistency,

    Y .-C. Chen, Y .-Y . Lin, M.-H. Yang, and J.-B. Huang, “Crdoco: Pixel- level domain transfer with cross-domain consistency,” in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition , 2019, pp. 1791–1800

    2019

  9. [12]

    Game theoretic analysis of road user safety scenarios involving autonomous vehicles,

    U. Michieli and L. Badia, “Game theoretic analysis of road user safety scenarios involving autonomous vehicles,” in 2018 IEEE 29th Annual International Symposium on Personal, Indoor and Mobile Radio Communications. IEEE, 2018, pp. 1377–1381

    2018

  10. [13]

    The Cityscapes dataset for semantic urban scene understanding,

    M. Cordts, M. Omran, S. Ramos, T. Rehfeld, M. Enzweiler, R. Be- nenson, U. Franke, S. Roth, and B. Schiele, “The Cityscapes dataset for semantic urban scene understanding,” in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition , 2016, pp. 3213–3223

    2016

  11. [14]

    The Mapillary vistas dataset for semantic understanding of street scenes,

    G. Neuhold, T. Ollmann, S. Rota Bulo, and P. Kontschieder, “The Mapillary vistas dataset for semantic understanding of street scenes,” in Proceedings of International Conference on Computer Vision , 2017, pp. 4990–4999

    2017

  12. [15]

    Bdd100k: A diverse driving video database with scalable annotation tooling,

    F. Yu, W. Xian, Y . Chen, F. Liu, M. Liao, V . Madhavan, and T. Darrell, “Bdd100k: A diverse driving video database with scalable annotation tooling,” arXiv preprint arXiv:1805.04687 , 2018

    2018 arXiv

  13. [17]

    Road: Reality oriented adaptation for semantic segmentation of urban scenes,

    Y . Chen, W. Li, and L. Van Gool, “Road: Reality oriented adaptation for semantic segmentation of urban scenes,” in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition , 2018, pp. 7892–7901

    2018

  14. [18]

    Attribute dissection of urban road scenes for effi- cient dataset integration,

    J. Kim and C. Park, “Attribute dissection of urban road scenes for effi- cient dataset integration,” in International Joint Conference on Artificial Intelligence Workshops, 2018, pp. 8–15

    2018

  15. [19]

    Semantic object classes in video: A high-definition ground truth database,

    G. Brostow, J. Fauqueur, and R. Cipolla, “Semantic object classes in video: A high-definition ground truth database,” Pattern Recognition Letters, pp. 88–97, 2009

    2009

  16. [20]

    Constrained convolutional neural networks for weakly supervised segmentation,

    D. Pathak, P. Krahenbuhl, and T. Darrell, “Constrained convolutional neural networks for weakly supervised segmentation,” in Proceedings of International Conference on Computer Vision , 2015, pp. 1796–1804

    2015

  17. [21]

    Semi and weakly supervised semantic segmentation using generative adversarial network,

    N. Souly, C. Spampinato, and M. Shah, “Semi and weakly supervised semantic segmentation using generative adversarial network,” arXiv preprint arXiv:1703.09695, 2017

    2017 arXiv

  18. [22]

    Towards weakly supervised seman- tic segmentation by means of multiple instance and multitask learning,

    A. Vezhnevets and J. M. Buhmann, “Towards weakly supervised seman- tic segmentation by means of multiple instance and multitask learning,” in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition. IEEE, 2010, pp. 3249–3256

    2010

  19. [23]

    STC: A simple to complex framework for weakly-supervised semantic segmentation,

    Y . Wei, X. Liang, Y . Chen, X. Shen, M.-M. Cheng, J. Feng, Y . Zhao, and S. Yan, “STC: A simple to complex framework for weakly-supervised semantic segmentation,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 39, no. 11, pp. 2314–2320, 2017

    2017

  20. [24]

    Decoupled deep neural network for semi- supervised semantic segmentation,

    S. Hong, H. Noh, and B. Han, “Decoupled deep neural network for semi- supervised semantic segmentation,” in Advances in Neural Information Processing Systems, 2015, pp. 1495–1503

    2015

  21. [25]

    Boxsup: Exploiting bounding boxes to super- vise convolutional networks for semantic segmentation,

    J. Dai, K. He, and J. Sun, “Boxsup: Exploiting bounding boxes to super- vise convolutional networks for semantic segmentation,” in Proceedings of International Conference on Computer Vision , 2015, pp. 1635–1643

    2015

  22. [26]

    Weakly-supervised semantic segmentation network with deep seeded region growing,

    Z. Huang, X. Wang, J. Wang, W. Liu, and J. Wang, “Weakly-supervised semantic segmentation network with deep seeded region growing,” in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, 2018, pp. 7014–7023

    2018

  23. [27]

    Saliency guided deep network for weakly-supervised image segmentation,

    F. Sun and W. Li, “Saliency guided deep network for weakly-supervised image segmentation,” Pattern Recognition Letters, 2019

    2019

  24. [28]

    Weakly-and semi-supervised learning of a deep convolutional network for semantic image segmentation,

    G. Papandreou, L.-C. Chen, K. P. Murphy, and A. L. Yuille, “Weakly-and semi-supervised learning of a deep convolutional network for semantic image segmentation,” in Proceedings of International Conference on Computer Vision, 2015, pp. 1742–1750

    2015

  25. [29]

    Learning from synthetic data: Addressing domain shift for semantic segmentation,

    S. Sankaranarayanan, Y . Balaji, A. Jain, S. Nam Lim, and R. Chellappa, “Learning from synthetic data: Addressing domain shift for semantic segmentation,” in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, 2018, pp. 3752–3761

    2018

  26. [30]

    Semantic segmentation using adversarial networks,

    P. Luc, C. Couprie, S. Chintala, and J. Verbeek, “Semantic segmentation using adversarial networks,” in NIPS Workshop on Adversarial Training, 2016

    2016

  27. [31]

    A deeper look at dataset bias,

    T. Tommasi, N. Patricia, B. Caputo, and T. Tuytelaars, “A deeper look at dataset bias,” in Domain Adaptation in Computer Vision Applications. Springer, 2017, pp. 37–55

    2017

  28. [32]

    Unbiased look at dataset bias,

    A. Torralba and A. Efros, “Unbiased look at dataset bias,” in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition. IEEE Computer Society, 2011, pp. 1521–1528

    2011

  29. [33]

    Overcoming dataset bias: An unsupervised domain adaptation approach,

    B. Gong, F. Sha, and K. Grauman, “Overcoming dataset bias: An unsupervised domain adaptation approach,” in NIPS Workshop on Large Scale Visual Recognition and Retrieval , vol. 3. Citeseer, 2012

    2012

  30. [34]

    Undoing the damage of dataset bias,

    A. Khosla, T. Zhou, T. Malisiewicz, A. A. Efros, and A. Torralba, “Undoing the damage of dataset bias,” in Proceedings of European Conference on Computer Vision . Springer, 2012, pp. 158–171

    2012

  31. [35]

    Learning from simulated and unsupervised images through adversarial training,

    A. Shrivastava, T. Pfister, O. Tuzel, J. Susskind, W. Wang, and R. Webb, “Learning from simulated and unsupervised images through adversarial training,” in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, 2017, pp. 2107–2116

    2017

  32. [36]

    Synthetic to real adaptation with generative correlation alignment networks,

    X. Peng and K. Saenko, “Synthetic to real adaptation with generative correlation alignment networks,” in 2018 IEEE Winter Conference on Applications of Computer Vision . IEEE, 2018, pp. 1982–1991

    2018

  33. [37]

    Generative image modeling using style and structure adversarial networks,

    X. Wang and A. Gupta, “Generative image modeling using style and structure adversarial networks,” in Proceedings of European Conference on Computer Vision . Springer, 2016, pp. 318–335

    2016

  34. [38]

    Generative visual manipulation on the natural image manifold,

    J.-Y . Zhu, P. Kr ¨ahenb¨uhl, E. Shechtman, and A. A. Efros, “Generative visual manipulation on the natural image manifold,” in Proceedings of European Conference on Computer Vision . Springer, 2016, pp. 597– 613. 11

    2016

  35. [39]

    Generating images with recurrent adversarial networks,

    D. J. Im, C. D. Kim, H. Jiang, and R. Memisevic, “Generating images with recurrent adversarial networks,” arXiv preprint arXiv:1602.05110 , 2016

    2016 arXiv

  36. [40]

    Weakly supervised object localization with progressive domain adaptation,

    D. Li, J.-B. Huang, Y . Li, S. Wang, and M.-H. Yang, “Weakly supervised object localization with progressive domain adaptation,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2016, pp. 3512–3520

    2016

  37. [41]

    Cross-domain weakly-supervised object detection through progressive domain adap- tation,

    N. Inoue, R. Furuta, T. Yamasaki, and K. Aizawa, “Cross-domain weakly-supervised object detection through progressive domain adap- tation,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 5001–5009

    2018

  38. [42]

    Unified deep supervised domain adaptation and generalization,

    S. Motiian, M. Piccirilli, D. A. Adjeroh, and G. Doretto, “Unified deep supervised domain adaptation and generalization,” in Proceedings of the IEEE International Conference on Computer Vision , 2017, pp. 5715– 5725

    2017

  39. [43]

    Semi- supervised domain adaptation via minimax entropy,

    K. Saito, D. Kim, S. Sclaroff, T. Darrell, and K. Saenko, “Semi- supervised domain adaptation via minimax entropy,” arXiv preprint arXiv:1904.06487, 2019

    1904 arXiv

  40. [44]

    Unsupervised domain adaptation by backpropagation,

    Y . Ganin and V . Lempitsky, “Unsupervised domain adaptation by backpropagation,” in International Conference on Machine Learning , 2015, pp. 1180–1189

    2015

  41. [45]

    Domain-adversarial training of neural networks,

    Y . Ganin, E. Ustinova, H. Ajakan, P. Germain, H. Larochelle, F. Lavio- lette, M. Marchand, and V . Lempitsky, “Domain-adversarial training of neural networks,” The Journal of Machine Learning Research , vol. 17, no. 1, pp. 2096–2030, 2016

    2016

  42. [46]

    Learning transferable features with deep adaptation networks,

    M. Long, Y . Cao, J. Wang, and M. Jordan, “Learning transferable features with deep adaptation networks,” in International Conference on Machine Learning , 2015, pp. 97–105

    2015

  43. [47]

    Simultaneous deep transfer across domains and tasks,

    E. Tzeng, J. Hoffman, T. Darrell, and K. Saenko, “Simultaneous deep transfer across domains and tasks,” in Proceedings of International Conference on Computer Vision , 2015, pp. 4068–4076

    2015

  44. [49]

    Unsupervised domain adaptation for tof data denoising with adversarial learning,

    G. Agresti, H. Schaefer, P. Sartor, and P. Zanuttigh, “Unsupervised domain adaptation for tof data denoising with adversarial learning,” in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, 2019, pp. 5584–5593

    2019

  45. [51]

    Taking a closer look at domain shift: Category-level adversaries for semantics consistent domain adaptation,

    Y . Luo, L. Zheng, T. Guan, J. Yu, and Y . Yang, “Taking a closer look at domain shift: Category-level adversaries for semantics consistent domain adaptation,” Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, 2019

    2019

  46. [52]

    Incremental Learning Techniques for Semantic Segmentation,

    U. Michieli and P. Zanuttigh, “Incremental Learning Techniques for Semantic Segmentation,” in Proceedings of International Conference on Computer Vision Workshops, 2019

    2019

  47. [54]

    Penalizing top performers: Conservative loss for semantic segmentation adaptation,

    X. Zhu, H. Zhou, C. Yang, J. Shi, and D. Lin, “Penalizing top performers: Conservative loss for semantic segmentation adaptation,” in Proceedings of European Conference on Computer Vision , 2018, pp. 568–583

    2018

  48. [55]

    Learning semantic segmentation from synthetic data: A geometrically guided input-output adaptation approach,

    Y . Chen, W. Li, X. Chen, and L. Van Gool, “Learning semantic segmentation from synthetic data: A geometrically guided input-output adaptation approach,” arXiv preprint arXiv:1812.05040 , 2018

    2018 arXiv

  49. [56]

    Fully convolutional adaptation networks for semantic segmentation,

    Y . Zhang, Z. Qiu, T. Yao, D. Liu, and T. Mei, “Fully convolutional adaptation networks for semantic segmentation,” in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition , 2018, pp. 6810–6818

    2018

  50. [58]

    Cycada: Cycle-consistent adversarial domain adapta- tion,

    J. Hoffman, E. Tzeng, T. Park, J.-Y . Zhu, P. Isola, K. Saenko, A. Efros, and T. Darrell, “Cycada: Cycle-consistent adversarial domain adapta- tion,” in Proceedings of the 35th International Conference on Machine Learning, 2018

    2018

  51. [59]

    Seeded region growing,

    R. Adams and L. Bischof, “Seeded region growing,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 16, no. 6, pp. 641– 647, 1994

    1994

  52. [60]

    Seednet: Automatic seed generation with deep reinforcement learning for robust interactive seg- mentation,

    G. Song, H. Myeong, and K. Mu Lee, “Seednet: Automatic seed generation with deep reinforcement learning for robust interactive seg- mentation,” in Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, 2018, pp. 1760–1768

    2018

  53. [61]

    Pre-computed weights for ResNet-101,

    V . Nekrasov, “Pre-computed weights for ResNet-101,” https://github.com/DrSleep/tensorflow-deeplab-resnet, Accessed: 2019-07-04

    2019

  54. [62]

    Microsoft coco: Common objects in context,

    T.-Y . Lin, M. Maire, S. Belongie, J. Hays, P. Perona, D. Ramanan, P. Doll ´ar, and C. L. Zitnick, “Microsoft coco: Common objects in context,” in Proceedings of European Conference on Computer Vision . Springer, 2014, pp. 740–755. Umberto Michieli received the M.Sc. degree in...

    2014

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.