Texture Underfitting for Domain Adaptation

Dengxin Dai; Jan-Nico Zaech; Luc Van Gool; Martin Hahner

REVIEW 3 major objections 4 minor 34 references

Texture Underfitting for Domain Adaptation

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 Randomly stylizing images with paintings and a two-stage training schedule makes neural segmentation models rely on shape rather than texture, improving synthetic-to-real domain adaptation.

desk verdict A simple, useful stylization trick for domain adaptation in segmentation, but the unsupervised evaluation protocol is post-hoc and needs fixing before the headline claim is solid. read the letter →

arxiv 1908.11215 v1 pith:3WULIUWW submitted 2019-08-29 cs.CV

Jan-Nico Zaech , Dengxin Dai , Martin Hahner , Luc Van Gool This is my paper · ORCID

classification cs.CV

keywords domainadaptationsemanticsegmentationtexturebiasimagestylizationadaptiveinstancenormalizationsynthetic-to-realtransferdeeplearningautonomousdriving

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 claims that domain adaptation for semantic segmentation can be improved by deliberately preventing networks from overfitting to texture. It does so by stylizing every training image with the style of a random painting, which randomizes object textures while largely keeping their shapes, and then training in two stages: first on a mix of conventional and stylized data, then on conventional data only. This forces the model to learn transferable shape information before it refines target-domain texture. Supporting experiments show gains in both unsupervised adaptation (33.93 versus 32.84 mean IoU) and supervised fine-tuning with very few labeled target images (39.65 versus 38.38 with 5 Cityscapes images). If true, the method offers a data-only way to boost any existing adaptation pipeline without changing the network.

What carries the argument

The load-bearing objects are the stylized dataset and the two-stage training schedule. Each image from the source and target datasets is restyled by adaptive instance normalization using the style of a random painting, so textures vary from image to image while content structure is preserved; training then alternates between a combined dataset (stylized and conventional examples drawn with equal probability) for pre-training and the conventional dataset for fine-tuning. This schedule is what lets the model first learn a texture-invariant, shape-based representation that is more similar between the two domains, and then pick up the target-domain textural cues needed for fine-grained boundaries.

What would settle it

Compare the combined-dataset pre-training schedule against a control where style transfer is replaced by a texture randomization that provably preserves edges, such as color-only jitter or noise confined to texture regions. If that control matches the 33.93 mIoU result, the mechanism is texture randomization, not stylization; if it falls back to the 32.84 baseline, the specific style-transfer perturbation is essential. A second check is to vary the pre-training/fine-tuning iteration split: if the gain disappears when the switch happens earlier or later, the two-stage schedule is the load-bearing component.

Watch

Extended reading notes

Core claim

The central discovery is that randomizing image textures at training time, and scheduling training so that shape-based pre-training precedes texture-based fine-tuning, improves transfer from synthetic driving scenes to real ones. On the synthetic-to-real semantic segmentation task, stylizing both source and target images with arbitrary painting styles destroys the class-texture correlation, forcing the network to rely on object boundaries and global structure—information that transfers across domains. The authors demonstrate this on the AdaptSegNet unsupervised pipeline (33.93 versus 32.84 mean IoU) and on supervised fine-tuning with 5, 10, and 20 labeled Cityscapes images, where pre-training on the combined stylized-plus-conventional data outperforms conventional GTA pre-training until the gap narrows at 20 images.

Load-bearing premise

The load-bearing premise is that random painting stylization destroys texture cues while leaving the object shapes and boundaries that segmentation depends on sufficiently intact; the authors themselves note that stylization inevitably perturbs local structure.

Editorial extensions

If this is right

Because stylization is a data-engineering step, it can be plugged into existing unsupervised or supervised domain-adaptation methods without modifying the architecture or loss functions.
The benefit is largest when target-domain labels are scarce: with 5 Cityscapes images, combined pre-training improves mIoU from 38.38 to 39.65, while the gain nearly vanishes at 20 images.
Pre-training on stylized data alone hurts (32.43 mIoU with 5 images), so the improvement depends on mixing conventional and stylized data in pre-training, not on using stylization exclusively.
Qualitative results indicate gains concentrate on well-defined shapes such as sidewalks and buildings and reduce small isolated mislabeled regions, suggesting the mechanism improves boundary awareness.

Reading between the lines

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

If the mechanism is texture randomization rather than the specific painting style, cheaper corruptions such as color jitter, channel shuffling, or noise confined to texture regions should produce comparable gains; the paper does not test this.
The near-tie at 20 Cityscapes images suggests that the value of stylization scales inversely with the amount of target-domain supervision, so the method may matter again for rare classes or imbalanced label distributions even when total labels are plentiful.
The same two-stage texture-underfitting recipe could transfer to other dense prediction tasks where shape dominates, such as depth estimation or instance segmentation, but generalization beyond semantic segmentation is not established.
Because style transfer perturbs local structure, a style pool curated to preserve edges, or an edge-aware regularization after stylization, might extend the gains; this is a testable design choice rather than a paper claim.

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

Two things to know. First, the core idea is genuinely simple and useful: they take the random style transfer that Geirhos used to break texture bias in classification, and apply it to synthetic-to-real semantic segmentation, with a two-stage training scheme (pre-train on stylized + real, fine-tune on real). It's a plug-in data-engineering trick that seems to help in both unsupervised and limited-supervision settings. Second, the headline unsupervised result is not as sturdy as the abstract implies, because the evaluation protocol optimizes the averaging window location separately for each method. That's a real flaw, but not a fatal one.

What's new: nobody had used random stylization for domain adaptation in segmentation before. The two-stage training is a reasonable response to the fact that stylization hurts segmentation when used alone. The paper is also honestly written: they run ten trials (more than most), and they explicitly say that stylized-only pretraining degrades performance and that the advantage vanishes at 20 labeled images in the supervised setting.

Where it's soft: in Section IV-A they average mIoU over a seven-step window and then "optimize its location for each method individually" — conventional gets 110k–135k, combined gets 125k–150k. That is post-hoc model selection on the validation curves, and it can inflate the 1.09 mIoU gap. They don't report standard deviations or any significance test, so we can't tell how much run-to-run noise matters. The Figure 3 curves do suggest the combined method is better over many iterations, not just at the chosen windows, so the effect is probably real, but the current numbers aren't a clean estimate of it.

The supervised tables are more trustworthy — the checkpoints are selected on a held-out set — and there the gain is about 1.3 mIoU at 5 shots, 0.5 at 10, and 0.01 at 20. That's an honest picture, and the paper says so.

Who it's for: anyone doing domain adaptation for segmentation, or studying texture bias in real tasks. It's not a deep theoretical contribution, but it's a practical result that others will want to build on. It deserves a serious referee. The evaluation flaw should be fixed — either use a fixed window/snapshot or report variance — but the central idea holds up. If the authors address the protocol, I'd be comfortable with a weak accept.

Referee Report

3 major / 4 minor

Summary. The paper proposes a data-augmentation and training procedure for domain adaptation in semantic segmentation, motivated by the texture bias of CNNs. The authors use AdaIN-based random image stylization to create texture-randomized versions of source and target images, then pre-train on the combined conventional plus stylized datasets and fine-tune on conventional data. They evaluate on GTA-to-Cityscapes transfer in two settings: unsupervised domain adaptation with AdaptSegNet and supervised domain adaptation by fine-tuning on 5, 10, or 20 labeled Cityscapes images. The central claim is that this texture-underfitting approach outperforms conventional training methods.

Significance. If the claims are sustained, the paper offers a simple, orthogonal plug-in strategy for domain adaptation that connects texture-bias findings in classification to dense prediction tasks. The experimental design has strengths: ten runs for the unsupervised comparison, use of public datasets, and a clear baseline comparison. However, the headline unsupervised result is weakened by a post-hoc evaluation protocol and missing variance reporting, and the supervised results are not consistently significant across data amounts. The paper would be a useful contribution if the evaluation is made fairer and the scope of the claims is made more precise.

major comments (3)

[IV-A, Figure 3, Table I] The unsupervised comparison is not made under a fixed evaluation protocol. The text states that the averaging window length is seven time-steps and that "we optimize its location for each method individually," with the conventional method evaluated at iterations 110k-135k and the combined method at 125k-150k. Selecting the window location on the same validation curves that are being compared can systematically inflate the reported 1.09 mIoU advantage (33.93 vs. 32.84). The stated rationale for window averaging is to approximate the practical setting where the exact stopping point is unknown, which argues for a fixed or pre-specified rule rather than per-method optimization. No standard deviations or significance tests are reported despite ten runs being performed, so the reader cannot assess run-to-run variability. Please re-evaluate both methods under a fixed checkpoint-selection policy (e.g., a pre-registered iteration or a fixed window at the end of training) and report error bars or pairwise significance information.
[IV-B, Table II] The supervised experiments do not consistently support the general claim that the proposed approach outperforms conventional training. At 20 Cityscapes images, CGTA (41.92 mIoU) is essentially tied with GTA pre-training (41.91 mIoU); at 10 images the gain is 0.51 mIoU; only at 5 images is the gain substantial (1.27 mIoU). The table reports a single checkpoint per setup, selected by best validation performance on a held-out set, with no multiple runs or significance testing. The claim in the abstract that the approach "outperforms conventional training methods" is therefore too broad. Please either restrict the claim to the low-data regime or provide statistical support such as multiple runs with confidence intervals, or a significance test across the different labeled-set sizes.
[V and Table II] The mechanism is not sufficiently isolated by the experiments. Stylized-only pre-training (SGTA) is substantially worse than conventional pre-training (e.g., 32.43 vs. 38.38 mIoU with 5 images), and the authors acknowledge in Section V that "image stylization inevitably perturbs local structure" and impedes segmentation. The success of the combined procedure may therefore hinge on the carefully chosen pre-training/fine-tuning split and on the sampling probability p=0.5, but no ablations are reported for these hyperparameters. Please add ablations varying p and the iteration split, and ideally compare against a control augmentation that is not texture-targeted, to support the interpretation that texture underfitting rather than generic data augmentation is responsible for the observed gains.

minor comments (4)

[Table I caption] The table caption says results are evaluated "over 10 runs and a window of 6 iterations," while Section IV-A says the window length is seven time-steps; please reconcile this discrepancy.
[Figure 3] The legend and caption contain typos: "Conventional datased" should be "Conventional dataset" and "Combined datased" should be "Combined dataset."
[III-C] There are several typos in this section: "disciminator" should be "discriminator," and "providing he final high-quality segmentation masks" should be "providing the final high-quality segmentation masks."
[IV-A] The text lists fine-tuning duration as "additional 75.000 iterations" with a period as thousands separator; this should be "75,000" or "75 000" for consistency with the rest of the paper.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the method is an empirical, externally benchmarked data-augmentation pipeline; no prediction reduces by construction to its inputs.

full rationale

The paper's central claim is an empirical comparison against external benchmarks (GTA to Cityscapes) using an external stylization method (AdaIN, [15]), an external baseline (AdaptSegNet, [28]), and an external texture-bias premise ([11], Geirhos et al.). No equation in the paper defines a quantity in terms of the quantity it is supposed to predict; no fitted parameter is renamed as a prediction. The two-stage training procedure (pre-train on combined stylized plus conventional data, then fine-tune on conventional data) is evaluated through measured mIoU on the Cityscapes validation and test sets, so the outcome is not imposed by construction. Self-citations in Section II-C ([5], [6], [26]) are background references to the authors' prior domain-adaptation work and are not load-bearing for the new claim. Section V explicitly acknowledges that stylization perturbs local structure and that stylized-only pre-training (SGTA) degrades performance (Table II, 32.43 vs 38.38 mIoU with 5 images); this is an honest limitation, not a circular step. The main caveat is the unsupervised evaluation protocol in Section IV-A: the averaging window is optimized separately for each method ('we optimize its location for each method individually'), and ten runs are reported without error bars. This weakens the strength of the headline gain but is a statistical rigor concern, not circularity: the reported numbers still come from the actual validation curves and are not algebraically forced by the method's definition. Overall, the derivation chain is self-contained against external evidence.

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

The central claim rests on hand-chosen training hyperparameters (pre-training/fine-tuning iteration split, dataset sampling probability, evaluation window) and on domain assumptions about texture bias, structural similarity across domains, and the structure-preserving property of stylization. No invented entities are introduced.

free parameters (4)

pre-training iterations = 85,000
Number of iterations for pre-training on the combined dataset before switching to fine-tuning; chosen by hand (Section IV-A).
fine-tuning iterations = 75,000
Number of iterations for fine-tuning on the conventional dataset after pre-training; set so the total equals the conventional baseline's 160,000 iterations (Section IV-A).
combined dataset sampling probability p = 0.5
Probability of drawing an image from the conventional vs stylized dataset during pre-training (Equation 1 and Section IV-A).
evaluation window length and location = length 7; conventional 110k-135k, combined 125k-150k
Averaging window for mIoU: length fixed at seven timesteps, location optimized per method on the validation set (Section IV-A). This is a hand-chosen evaluation parameter that materially affects reported results.

assumptions (4)

domain assumption Segmentation maps share structural similarities across source and target domains.
Stated in the Abstract and Introduction ('Under the assumption of structural similarities between segmentation maps, domain adaptation promises to resolve this challenge'). The whole method is premised on this transferability.
domain assumption In synthetic-to-real domain adaptation, structures (shapes) are more consistent across domains than textures.
Section I: 'We observe that in the context of synthetic-to-real domain adaptation, structures (shapes) are more consistent across the two domains than textures.' This observation motivates texture underfitting; there is no direct measurement supporting it beyond the final results.
domain assumption Random image stylization randomizes textures while preserving structures sufficiently for segmentation.
Section III-A: 'the stylization step randomizes the textures of the images while preserving the structures.' The Discussion qualifies this ('image stylization inevitably perturbs local structure'), so the sufficient preservation is an unproved assumption that the two-stage training is designed to compensate for.
domain assumption Neural networks are biased towards texture over structure.
Borrowed from Geirhos et al. [11], cited in Section II-D. The paper's entire motivation rests on this prior empirical finding; if the bias were absent, texture randomization would not change learning behavior.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Texture Underfitting for Domain Adaptation." pith.science (2026). https://pith.science/paper/3WULIUWW

@misc{pith2026190811215,
  author       = {Pith},
  title        = {Pith review of: Texture Underfitting for Domain Adaptation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3WULIUWW}},
  note         = {Machine review of arXiv:1908.11215}
}

read the original abstract

Comprehensive semantic segmentation is one of the key components for robust scene understanding and a requirement to enable autonomous driving. Driven by large scale datasets, convolutional neural networks show impressive results on this task. However, a segmentation algorithm generalizing to various scenes and conditions would require an enormously diverse dataset, making the labour intensive data acquisition and labeling process prohibitively expensive. Under the assumption of structural similarities between segmentation maps, domain adaptation promises to resolve this challenge by transferring knowledge from existing, potentially simulated datasets to new environments where no supervision exists. While the performance of this approach is contingent on the concept that neural networks learn a high level understanding of scene structure, recent work suggests that neural networks are biased towards overfitting to texture instead of learning structural and shape information. Considering the ideas underlying semantic segmentation, we employ random image stylization to augment the training dataset and propose a training procedure that facilitates texture underfitting to improve the performance of domain adaptation. In experiments with supervised as well as unsupervised methods for the task of synthetic-to-real domain adaptation, we show that our approach outperforms conventional training methods.

Figures

Figures reproduced from arXiv: 1908.11215 by the authors.

**Figure 3.** Average mIoU on the Cityscapes validation set. Plots [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗

**Figure 4.** Comparison of qualitative results. All frameworks [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

34 extracted references · 26 canonical work pages

[1]

Approximating CNNs with bag-of-local- features models works surprisingly well on imagenet,

W. Brendel and M. Bethge, “Approximating CNNs with bag-of-local- features models works surprisingly well on imagenet,” in International Conference on Learning Representations , 2019

work page 2019
[2]

DeepLab: Semantic Image Segmentation with Deep Con- volutional 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 Con- volutional Nets, Atrous Convolution, and Fully Connected CRFs,” arXiv:1606.00915 [cs] , June 2016

arXiv 2016
[3]

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 CVPR, 2018

work page 2018
[4]

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 CVPR, 2016

work page 2016
[5]

Model adaptation with synthetic and real data for semantic dense foggy scene under- standing,

D. Dai, C. Sakaridis, S. Hecker, and L. Van Gool, “Model adaptation with synthetic and real data for semantic dense foggy scene under- standing,” International Journal of Computer Vision (IJCV) , 2019

work page 2019
[6]

Dark model adaptation: Semantic image segmentation from daytime to nighttime,

D. Dai and L. Van Gool, “Dark model adaptation: Semantic image segmentation from daytime to nighttime,” in IEEE International Conference on Intelligent Transportation Systems , 2018

work page 2018
[7]

CARLA: An Open Urban Driving Simulator,

A. Dosovitskiy, G. Ros, F. Codevilla, A. Lopez, and V . Koltun, “CARLA: An Open Urban Driving Simulator,” in Annual Conference on Robot Learning , vol. 78, 2017

work page 2017
[8]

Unsupervised Domain Adaptation by Backpropagation,

Y . Ganin and V . Lempitsky, “Unsupervised Domain Adaptation by Backpropagation,” in Proceedings of International Conference on International Conference on Machine Learning , 2015

work page 2015

Show all 34 references

[9]

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,” arXiv:1505.07818 [cs, stat] , May 2015

2015 arXiv
[10]

Texture and art with deep neural networks,

L. A. Gatys, A. S. Ecker, and M. Bethge, “Texture and art with deep neural networks,” Current Opinion in Neurobiology , vol. 46, pp. 178– 186, Oct. 2017

2017
[11]

ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustness,

R. Geirhos, P. Rubisch, C. Michaelis, M. Bethge, F. A. Wich- mann, and W. Brendel, “ImageNet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustness,” arXiv:1811.12231 [cs, q-bio, stat] , Nov. 2018

2018 arXiv
[12]

Spatial Pyramid Pooling in Deep Convolutional Networks for Visual Recognition,

K. He, X. Zhang, S. Ren, and J. Sun, “Spatial Pyramid Pooling in Deep Convolutional Networks for Visual Recognition,” arXiv:1406.4729 [cs], vol. 8691, pp. 346–361, 2014

2014 arXiv
[13]

Deep residual learning for image recognition,

——, “Deep residual learning for image recognition,” in IEEE Con- ference on Computer Vision and Pattern Recognition , 2016

2016
[14]

CyCADA: Cycle-consistent adversarial domain adap- tation,

J. Hoffman, E. Tzeng, T. Park, J.-Y . Zhu, P. Isola, K. Saenko, A. Efros, and T. Darrell, “CyCADA: Cycle-consistent adversarial domain adap- tation,” in International Conference on Machine Learning , 2018

2018
[15]

Arbitrary Style Transfer in Real-Time with Adaptive Instance Normalization,

X. Huang and S. Belongie, “Arbitrary Style Transfer in Real-Time with Adaptive Instance Normalization,” in IEEE International Conference on Computer Vision (ICCV) , 2017

2017
[16]

Deep Neural Networks: A New Framework for Mod- eling Biological Vision and Brain Information Processing,

N. Kriegeskorte, “Deep Neural Networks: A New Framework for Mod- eling Biological Vision and Brain Information Processing,” Annual Review of Vision Science , vol. 1, no. 1, pp. 417–446, 2015

2015
[17]

Deep learning,

Y . LeCun, Y . Bengio, and G. Hinton, “Deep learning,” Nature, vol. 521, no. 7553, pp. 436–444, May 2015

2015
[18]

Fully convolutional networks for semantic segmentation,

J. Long, E. Shelhamer, and T. Darrell, “Fully convolutional networks for semantic segmentation,” in CVPR, 2015

2015
[19]

The Map- illary Vistas Dataset for Semantic Understanding of Street Scenes,

G. Neuhold, T. Ollmann, S. R. Bulo, and P. Kontschieder, “The Map- illary Vistas Dataset for Semantic Understanding of Street Scenes,” in 2017 IEEE International Conference on Computer Vision (ICCV) . Venice: IEEE, Oct. 2017, pp. 5000–5009

2017
[20]

A Survey on Transfer Learning,

S. J. Pan and Q. Yang, “A Survey on Transfer Learning,” IEEE Transactions on Knowledge and Data Engineering , vol. 22, no. 10, pp. 1345–1359, Oct. 2010

2010
[21]

Playing for Benchmarks,

S. R. Richter, Z. Hayder, and V . Koltun, “Playing for Benchmarks,” in IEEE International Conference on Computer Vision (ICCV) , 2017

2017
[22]

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 Computer Vision ECCV 2016, B. Leibe, J. Matas, N. Sebe, and M. Welling, Eds. Cham: Springer International Publishing, 2016, vol. 9906, pp. 102–118

2016
[23]

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 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) . Las Vegas, NV , USA: IEEE, June...

2016
[24]

ImageNet Large Scale Visual Recognition Challenge,

O. Russakovsky, J. Deng, H. Su, J. Krause, S. Satheesh, S. Ma, Z. Huang, A. Karpathy, A. Khosla, M. Bernstein, A. C. Berg, and L. Fei-Fei, “ImageNet Large Scale Visual Recognition Challenge,” International Journal of Computer Vision (IJCV) , vol. 115, no. 3, pp. 211–252, 2015

2015
[25]

Semantic foggy scene understanding with synthetic data,

C. Sakaridis, D. Dai, and L. Van Gool, “Semantic foggy scene understanding with synthetic data,” International Journal of Computer Vision, 2018

2018
[26]

Guided curriculum model adaptation and uncertainty-aware evaluation for semantic nighttime image segmentation,

——, “Guided curriculum model adaptation and uncertainty-aware evaluation for semantic nighttime image segmentation,” in Interna- tional Conference on Computer Vision (ICCV) , 2019

2019
[27]

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 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , 2017

2017
[28]

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 IEEE Conference on Computer Vision and Pattern Recognition, 2018

2018
[29]

Domain Adap- tation for Structured Output via Discriminative Patch Representations,

Y .-H. Tsai, K. Sohn, S. Schulter, and M. Chandraker, “Domain Adap- tation for Structured Output via Discriminative Patch Representations,” arXiv:1901.05427 [cs] , Jan. 2019

1901 arXiv
[30]

Simultaneous deep transfer across domains and tasks,

E. Tzeng, J. Hoffman, T. Darrell, and K. Saenko, “Simultaneous deep transfer across domains and tasks,” in IEEE International Conference on Computer Vision (ICCV) , 2015

2015
[31]

Deep visual domain adaptation: A survey,

M. Wang and W. Deng, “Deep visual domain adaptation: A survey,” Neurocomputing, vol. 312, pp. 135–153, Oct. 2018

2018
[32]

Synscapes: A Photorealistic Synthetic Dataset for Street Scene Parsing,

M. Wrenninge and J. Unger, “Synscapes: A Photorealistic Synthetic Dataset for Street Scene Parsing,” arXiv:1810.08705 [cs] , Oct. 2018

2018 arXiv
[33]

Multi-Scale Context Aggregation by Dilated Convolutions,

F. Yu and V . Koltun, “Multi-Scale Context Aggregation by Dilated Convolutions,” in ICLR, 2016

2016
[34]

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 IEEE International Conferenceon Computer Vision (ICCV) , 2017

2017

Pith tools

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

[1] [1]

Approximating CNNs with bag-of-local- features models works surprisingly well on imagenet,

W. Brendel and M. Bethge, “Approximating CNNs with bag-of-local- features models works surprisingly well on imagenet,” in International Conference on Learning Representations , 2019

work page 2019

[2] [2]

DeepLab: Semantic Image Segmentation with Deep Con- volutional 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 Con- volutional Nets, Atrous Convolution, and Fully Connected CRFs,” arXiv:1606.00915 [cs] , June 2016

arXiv 2016

[3] [3]

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 CVPR, 2018

work page 2018

[4] [4]

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 CVPR, 2016

work page 2016

[5] [5]

Model adaptation with synthetic and real data for semantic dense foggy scene under- standing,

D. Dai, C. Sakaridis, S. Hecker, and L. Van Gool, “Model adaptation with synthetic and real data for semantic dense foggy scene under- standing,” International Journal of Computer Vision (IJCV) , 2019

work page 2019

[6] [6]

Dark model adaptation: Semantic image segmentation from daytime to nighttime,

D. Dai and L. Van Gool, “Dark model adaptation: Semantic image segmentation from daytime to nighttime,” in IEEE International Conference on Intelligent Transportation Systems , 2018

work page 2018

[7] [7]

CARLA: An Open Urban Driving Simulator,

A. Dosovitskiy, G. Ros, F. Codevilla, A. Lopez, and V . Koltun, “CARLA: An Open Urban Driving Simulator,” in Annual Conference on Robot Learning , vol. 78, 2017

work page 2017

[8] [8]

Unsupervised Domain Adaptation by Backpropagation,

Y . Ganin and V . Lempitsky, “Unsupervised Domain Adaptation by Backpropagation,” in Proceedings of International Conference on International Conference on Machine Learning , 2015

work page 2015