Pith. sign in

REVIEW 4 major objections 5 minor 65 references

MaskAdapt: Unsupervised Geometry-Aware Domain Adaptation Using Multimodal Contextual Learning and RGB-Depth Masking

T0 review · 4 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read Depth-gradient cross-attention plus geometry-aware RGB-depth masking lets crop-weed segmentation adapt across fields, improving mean IoU by 0.99 to 10.86 points over the masked-image-consistency baseline on eight source-target pairs.

desk verdict Plausible engineering combo, but the SOTA claim is unverified: a possibly weakened MIC baseline, missing error bars, inconsistent prose gains, and one likely fabricated reference. read the letter →

arxiv 2505.24026 v1 pith:PUWRIBPT submitted 2025-05-29 cs.CV cs.AI

Numair Nadeem , Muhammad Hamza Asad , Saeed Anwar , Abdul Bais This is my paper
classification cs.CVcs.AI
keywords unsuperviseddomainadaptationsemanticsegmentationcrop-weedRGB-Dfusioncross-attentiondepthgradientsmaskedimagemodelingagriculturalrobotics
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 is trying to establish that an unsupervised domain adaptation pipeline for crop-weed semantic segmentation can generalize to new, unlabeled fields if it fuses RGB with depth-derived geometry through a depth-gradient-guided cross-attention module and trains with a geometry-aware complementary masking schedule. It argues that the key signal RGB-only UDA misses is the spatial transition in depth, gradual inside crop rows and sharp at crop-weed and crop-soil interfaces, and that this signal refines RGB features where crops and weeds overlap visually. On eight source-target pairs spanning two robotic platforms, multiple crop types, and different growth years, MaskAdapt reports the highest mean IoU among the compared methods, with gains over the masked-image-consistency baseline from +0.99 to +10.86 points and a top result of 88.58 percent on Maize WeedElec-to-BIPBIP. If correct, precision-agriculture systems can adapt segmentation to new fields without collecting new pixel labels, using only RGB images and depth maps estimated from them.

What carries the argument

The load-bearing mechanism is the depth-gradient-guided cross-attention module: for each encoder level $i$, depth features $F^{(i)}_{\mathrm{depth}}$ and the gradient magnitude $G^{(i)}_{\mathrm{depth}}$ are concatenated as $F^{(i)}_{\mathrm{depth+grad}}$, bilinearly downsampled, and linearly projected into queries $Q^{(i)}$ and keys $K^{(i)}$, while the RGB features become values $V^{(i)}$. The attention output is upsampled and added to the RGB features with a $1\times1$ convolution and residual connection, so depth geometry determines which RGB context aggregates. The second component is complementary geometry-aware masking: horizontal masks occlude full crop rows, vertical masks occlude inter-row soil bands, stochastic masks break local details, and the RGB and depth masks are complementary ($M_{\mathrm{depth}} = 1 - M_{\mathrm{rgb}}$) with a time-dependent masking ratio. Source-domain masking gives way to target-domain masking once pseudo-label confidence passes 90 percent, forcing the student model to infer occluded content from the surviving modality and from global context.

What would settle it

Retrain MaskAdapt on the same eight source-target pairs after replacing the estimated depth maps with corrupted versions, such as constant depth, spatially shifted depth, or noise with the same distribution, and compare mIOU. If the 88.58 percent result on Maize WeedElec-to-BIPBIP survives corruption, the depth-gradient cross-attention is not the mechanism; if performance collapses, the central claim depends on monocular depth being accurate enough to mark boundaries.

Watch

Extended reading notes

Core claim

MaskAdapt's central claim is that depth gradients, first-order spatial differences of the depth map, carry a geometric boundary cue that RGB texture alone cannot supply. Depth maps are produced by a pre-trained monocular Vision Transformer [39], and a lightweight depth encoder runs alongside a frozen pre-trained RGB encoder. In the fusion module, depth features and their gradient magnitude are concatenated, downsampled, and projected into queries and keys, while RGB features serve as values; the refined RGB feature keeps appearance details but aggregates them according to depth-similar regions, sharpening crop-weed and crop-soil boundaries. Around this, the paper wraps a masking schedule that corrupts RGB and depth complementarily with horizontal, vertical, or stochastic masks, first on the source domain and then on the target once pseudo-label confidence crosses 90 percent, forcing the model to reconstruct missing details from cross-modal and broader spatial context. The experimental claim is that this recipe, as a plug-in on top of the MIC student-teacher baseline, improves mean IoU on every evaluated pair, with weed and crop IoU rising by as much as 11.63 and 5.72 points in some shifts and a best overall mIOU of 88.58 percent.

Load-bearing premise

The load-bearing premise is that depth maps predicted from the same RGB images by a pre-trained monocular network are accurate enough that their gradients mark real crop-weed-soil boundaries; if those depth estimates are noisy or spatially misaligned, the cross-attention could blur RGB features and the reported gains would come from the masking strategies alone.

Editorial extensions

If this is right

  • Practitioners get a plug-in: MaskAdapt is built on the MIC student-teacher baseline and improves it on all eight source-target pairs, including cross-robot, cross-year, and cross-growth-stage shifts, without needing target labels.
  • Depth-gradient fusion has its own measurable effect: in the ablation, adding cross-attention with depth gradients raises mIOU from 78.98 to 81.35 percent on Maize WeedElec-to-BIPBIP, beyond the 79.46 percent for cross-attention without gradients.
  • The masking design carries most of the end-to-end gain: moving from stochastic-RGB-only masking to complementary geometry-aware masking with a dynamic ratio raises mIOU from 78.98 to 88.58 percent on the same pair.
  • The source-to-target masking schedule outperforms masking only the source or only the target, so there is a concrete training recipe for reducing pseudo-label noise while still adapting.

Reading between the lines

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

  • A testable extension the paper does not run is to replace the monocular depth estimates with ground-truth depth (stereo or LiDAR) and, separately, with deliberately corrupted depth; the difference would separate how much of the gain is true geometry from how much is an artifact of the depth estimator.
  • If depth-gradient guidance is as general as the ablations suggest, the same recipe could transfer to other row-structured domains such as orchards, vineyards, or road lanes, where a monocular depth prior is stable; the paper tests only maize and bean fields.
  • The mask orientations are fixed to three strategies; an adaptive variant that estimates the crop-row angle from the depth gradient and orients masks accordingly would test how much the method depends on row orientation being aligned with the image axes.
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

4 major / 5 minor

Summary. The paper proposes MaskAdapt, an unsupervised domain adaptation method for crop/weed semantic segmentation. It combines a frozen pre-trained RGB encoder with a lightweight depth encoder, computes depth gradients from monocularly estimated depth maps, fuses them with RGB features via a depth-gradient-guided cross-attention module, and applies horizontal/vertical/stochastic complementary masking with a source-to-target schedule based on pseudo-label confidence. Experiments on ROSE maize/bean datasets across six source-target pairs report higher mIOU than several UDA baselines, including MIC, with the largest gain on Maize WeedElec to BIPBIP (88.58 mIOU).

Significance. If the reported results are reproducible, MaskAdapt would be a practically useful plug-in for crop/weed UDA: Table 4 shows a +1.89 mIOU gain from adding depth gradients to cross-attention, and Table 3 shows a +0.67 mIOU gain from the scheduled masking scheme over target-only masking. The ablations are clearly decomposed and include seed-level variance on one adaptation pair, which is a strength. However, the headline claim of consistent state-of-the-art performance across all six pairs rests on Table 1, which lacks error bars and uses a possibly weakened MIC baseline; the prose gain ranges are also internally inconsistent. The contribution is therefore promising but not yet verified at the claimed level.

major comments (4)
  1. [Section 4.2, Table 1] Table 1 reports the central SOTA comparison with no variance estimates, although the caption says results are averaged over 3 random data sampling seeds and the ablation tables (Tables 2-4) report standard deviations of 0.10-0.30. Several claimed margins over MIC are small (e.g., +0.99 for Bean 2021 to 2019 and +1.43 for Bean 2019 to 2021), so without error bars or per-seed results the reader cannot tell whether these differences are significant. Please report mean plus/minus standard deviation (or confidence intervals) for every entry in Table 1, or at least for the MaskAdapt and baseline columns, and state the number of seeds used for each method.
  2. [Section 4.3.1, Table 2; Section 4.1 Training] The MIC [25] entry in Table 1 for Maize WeedElec to BIPBIP (78.98) is numerically identical to the Baseline (Stochastic) in Table 2 (78.98 plus/minus 0.20). This suggests that the MIC baseline used for comparison may be a stochastic-masking RGB baseline rather than the published MIC method, which uses an EMA teacher and masked-image consistency pseudo-labels. If so, the gains over MIC in Table 1 are not gains over the actual state of the art, and the main claim is inflated. Please specify the exact MIC reproduction (which components are included) and, if possible, compare against the official MIC model or report the exact configuration used.
  3. [Section 4.2] The prose reporting gains is internally inconsistent with Table 1. The text states the maximum mIOU improvement over the best existing method is 4.47% on Maize WeedElec to BIPBIP, but the gain over the MIC baseline in Table 1 is 9.60 points (78.98 to 88.58); it later states a minimum of 0.95% on Bean 2021 to 2019 while Table 1 gives 0.99, and a maximum of 5.19% on Maize 2021 to 2019 while Table 1 gives 10.86 over MIC. The sentences need to identify the comparison baseline consistently (MIC vs. best non-MIC) and correct the numbers to match Table 1.
  4. [Section 4.1 Datasets and Training] The paper states that the target domain is split into training/validation and that performance is evaluated on a labeled holdout validation, but it does not say whether the 90% confidence threshold, the masking-ratio schedule m_t, or the pooling factors p_i were tuned on that validation set. If these were selected on the labeled target split, the method is not a clean unsupervised adaptation, and the comparison with methods whose hyperparameters were not tuned on the same target split would be unfair. Please clarify the model-selection protocol and provide a sensitivity analysis for these hyperparameters.
minor comments (5)
  1. [Section 4.2] The phrase 'existing best methods' is used ambiguously; in the same paragraph the improvement ranges switch between comparing to MIC and comparing to another method. Use a single, explicitly named comparator throughout.
  2. [Section 4.1 Depth Images] The source of the monocular depth maps is named only as 'Vision Transformer [39]'; specify the exact pre-trained model and any preprocessing (e.g., resize, alignment) so the experiments are reproducible, and ideally show a qualitative comparison of estimated depth gradients against RGB boundaries.
  3. [Table 1] The table is very hard to read because numeric columns appear concatenated without spacing; reformat with proper column separation and align the classwise IOUs under each method.
  4. [Section 3.2] The masking formulation defines M_Theta,rgb(a,b) using a block index (a,b) but does not specify the block size or how block locations are sampled; add this detail.
  5. [References] Reference [4] appears to be a technical report with no arXiv/DOI and generic author names; please verify the reference and provide a permanent identifier if it is real.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: the central mIOU claims are empirical benchmark measurements on a labeled target holdout, not derivations from the method's own outputs; only minor self-citations and a baseline-identity caveat warrant a low caution flag.

full rationale

The paper's central claim (consistently outperforming SOTA UDA methods in mIOU) is an empirical benchmark result evaluated on a labeled target-domain holdout, as stated in Section 3 ('performance evaluated on a set of labeled holdout validation of the target domain') and Section 4.1. The proposed depth gradient feature is defined operationally from DPT-predicted depth via first-order differences (Eq. 1) and is not a restatement of the target mIOU; the cross-attention (Eqs. 2-6) and masking schedule are trained with standard cross-entropy and consistency losses. None of the paper's equations reduce a predicted quantity to a fitted input, and no central premise depends on a uniqueness theorem or on an ansatz justified only by self-citation. The self-citations [1], [2], [28], and [37] support auxiliary context or are baseline comparisons and are not load-bearing. Two reporting issues (the MIC column in Table 1 numerically matches the stochastic-RGB ablation baseline in Table 2, and the prose gain ranges in Section 4.2 do not match Table 1) are reproducibility and fairness risks, but they are not examples of a derivation being equivalent to its inputs by construction, so they do not raise the circularity score beyond a low caution level.

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

The paper contributes no new theoretical constructs; it relies on a set of hand-chosen hyperparameters (masking schedule, confidence threshold, pooling factors) and on the assumption that monocular depth estimates carry real geometric signal. These are the main non-independent elements of the method.

free parameters (3)
  • masking_ratio_schedule_mt = approximately 10-20% at start, rising to 80%
    The dynamic masking ratio in Section 3.2 is described as low early and high later, but the exact schedule function is not given, and its curve is a hand-chosen component of the method that could be tuned to the target split.
  • pseudo_label_confidence_threshold = 0.90
    Threshold in Section 4.1 Training for switching from source masking to target masking; no sensitivity analysis is provided and the value is likely selected to maximize validation mIOU.
  • attention_pooling_factors_pi = not reported
    The per-level downsampling factors in Equations (2) to (4) control the spatial resolution of cross-attention and are not specified in the paper, which is a replication gap.
assumptions (3)
  • domain assumption DPT monocular depth estimates are accurate enough to recover crop-weed-soil boundaries for the considered datasets.
    All depth maps are estimated from RGB via a pretrained Vision Transformer (Section 4.1 Depth Images); the paper does not validate depth quality against ground-truth geometry.
  • domain assumption Crop rows in the images are predominantly horizontal or vertical, so fixed mask orientations align with the scene structure.
    The horizontal, vertical, and stochastic masking in Section 3.2 assumes a known row orientation; no rotation-invariant treatment is provided.
  • ad hoc to paper Pseudo-labels from the EMA teacher are reliable once confidence exceeds 90%.
    The 90% threshold introduced in Section 4.1 gates target-domain masking; its accuracy is not analyzed and it may not transfer to other datasets.

how reviews work

0 comments
Cite this review

Pith. "Pith review of MaskAdapt: Unsupervised Geometry-Aware Domain Adaptation Using Multimodal Contextual Learning and RGB-Depth Masking." pith.science (2026). https://pith.science/paper/PUWRIBPT

@misc{pith2026250524026,
  author       = {Pith},
  title        = {Pith review of: MaskAdapt: Unsupervised Geometry-Aware Domain Adaptation Using Multimodal Contextual Learning and RGB-Depth Masking},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PUWRIBPT}},
  note         = {Machine review of arXiv:2505.24026}
}
read the original abstract

Semantic segmentation of crops and weeds is crucial for site-specific farm management; however, most existing methods depend on labor intensive pixel-level annotations. A further challenge arises when models trained on one field (source domain) fail to generalize to new fields (target domain) due to domain shifts, such as variations in lighting, camera setups, soil composition, and crop growth stages. Unsupervised Domain Adaptation (UDA) addresses this by enabling adaptation without target-domain labels, but current UDA methods struggle with occlusions and visual blending between crops and weeds, leading to misclassifications in real-world conditions. To overcome these limitations, we introduce MaskAdapt, a novel approach that enhances segmentation accuracy through multimodal contextual learning by integrating RGB images with features derived from depth data. By computing depth gradients from depth maps, our method captures spatial transitions that help resolve texture ambiguities. These gradients, through a cross-attention mechanism, refines RGB feature representations, resulting in sharper boundary delineation. In addition, we propose a geometry-aware masking strategy that applies horizontal, vertical, and stochastic masks during training. This encourages the model to focus on the broader spatial context for robust visual recognition. Evaluations on real agricultural datasets demonstrate that MaskAdapt consistently outperforms existing State-of-the-Art (SOTA) UDA methods, achieving improved segmentation mean Intersection over Union (mIOU) across diverse field conditions.

Figures

Figures reproduced from arXiv: 2505.24026 by the authors.

Figure 1
Figure 1. Performance comparison of our proposed method with [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 3
Figure 3. Illustration of our training pipeline featuring Multimodal [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figure 4
Figure 4. Our Depth Gradient-Guided Cross-Attention mecha [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗
Figures from the paper (1 more)
Figure 5
Figure 5. Figure 5: Qualitative results are shown for Bean BIPBIP [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

65 extracted references · 59 canonical work pages

  1. [4]

    Geometry-aware masking strategies for crop row analysis

    John Chen and Robert Smith. Geometry-aware masking strategies for crop row analysis. Technical Report MITB3- TR-2025-01, Massachusetts Institute of Technology, Depart- ment of Electrical Engineering and Computer Science, Cam- bridge, MA, 2025. Technical report detailing vertical and horizontal masking techniques for agricultural imaging. 5

    work page 2025

  2. [25]

    MIC: Masked image consistency for context- enhanced domain adaptation

    Lukas Hoyer, Dengxin Dai, Haoran Wang, and Luc Van Gool. MIC: Masked image consistency for context- enhanced domain adaptation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2023. 2, 3, 4, 5, 6, 7

    work page 2023

  3. [1]

    Weed detection in canola fields using maximum likelihood classification and deep convolutional neural network

    Muhammad Hamza Asad and Abdul Bais. Weed detection in canola fields using maximum likelihood classification and deep convolutional neural network. Information Processing in Agriculture, 7(4):535–545, 2020. 1

    work page 2020

  4. [2]

    Improved crop and weed detection with diverse data ensem- ble learning

    Muhammad Hamza Asad, Saeed Anwar, and Abdul Bais. Improved crop and weed detection with diverse data ensem- ble learning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 5336– 5345, 2024. 1, 5

    work page 2024

  5. [3]

    A comparative study of fourier trans- form and cyclegan as domain adaptation techniques for weed segmentation - code and data, 2023

    Riccardo Bertoglio, Alessio Mazzucchelli, Nico Catalano, and Matteo Matteucci. A comparative study of fourier trans- form and cyclegan as domain adaptation techniques for weed segmentation - code and data, 2023. 1, 5

    work page 2023

  6. [5]

    3d sketch-aware semantic scene comple- tion via semi-supervised structure prior

    Xiaokang Chen, Kwan-Yee Lin, Chen Qian, Gang Zeng, and Hongdong Li. 3d sketch-aware semantic scene comple- tion via semi-supervised structure prior. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020. 2

    work page 2020

  7. [6]

    Bi-directional cross-modality feature propagation with separation-and- aggregation gate for rgb-d semantic segmentation

    Xiaokang Chen, Kwan-Yee Lin, Jingdong Wang, Wayne Wu, Chen Qian, Hongdong Li, and Gang Zeng. Bi-directional cross-modality feature propagation with separation-and- aggregation gate for rgb-d semantic segmentation. In Euro- pean Conference on Computer Vision (ECCV) , pages 561– 577, 2020. 2

    work page 2020

  8. [7]

    Learning semantic segmentation from synthetic data: A geo- metrically guided input-output adaptation approach

    Yuhua Chen, Wen Li, Xiaoran Chen, and Luc Van Gool. Learning semantic segmentation from synthetic data: A geo- metrically guided input-output adaptation approach. In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 1841–1850, 2019. 3

    work page 2019

Show all 65 references
  1. [8]

    Scale-aware domain adap- tive faster r-cnn

    Yuhua Chen, Haoran Wang, Wen Li, Christos Sakaridis, Dengxin Dai, and Luc Van Gool. Scale-aware domain adap- tive faster r-cnn. International Journal of Computer Vision (IJCV), 129(7):2223–2243, 2021. 3, 4

    2021

  2. [9]

    Locality-sensitive deconvolution networks with gated fusion for rgb-d indoor semantic segmentation

    Yong Cheng, Rui Cai, Zhi Li, Xin Zhao, and Kaiqi Huang. Locality-sensitive deconvolution networks with gated fusion for rgb-d indoor semantic segmentation. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 3029–3037, 2017. 2

    2017

  3. [10]

    Self- ensembling with gan-based data augmentation for domain adaptation in semantic segmentation

    Jaehoon Choi, Taekyung Kim, and Changick Kim. Self- ensembling with gan-based data augmentation for domain adaptation in semantic segmentation. In ICCV, pages 6830– 6840, 2019. 3

    2019

  4. [11]

    MMSegmentation: Openmmlab semantic segmentation toolbox and benchmark

    MMSegmentation Contributors. MMSegmentation: Openmmlab semantic segmentation toolbox and benchmark. https : / / github . com / open - mmlab/mmsegmentation, 2020. 5

    2020

  5. [12]

    Peco: Perceptual codebook for bert pre-training of vision transformers

    Xiaoyi Dong, Jianmin Bao, Ting Zhang, Dongdong Chen, Weiming Zhang, Lu Yuan, Dong Chen, Fang Wen, and Nenghai Yu. Peco: Perceptual codebook for bert pre-training of vision transformers. arXiv preprint arXiv:2111.12710 ,

    arXiv

  6. [13]

    An image is worth 16x16 words: Transformers for image recognition at scale

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Syl- vain Gelly, Nicolas Usunier, and Neil Houlsby. An image is worth 16x16 words: Transformers for image recognition ...

    2021

  7. [14]

    Seed vigour and crop establishment: extending performance be- yond adaptation

    William E Finch-Savage and George W Bassel. Seed vigour and crop establishment: extending performance be- yond adaptation. Journal of experimental botany, 67(3):567– 591, 2016. 1

    2016

  8. [15]

    Domain-adversarial train- ing of neural networks

    Yaroslav Ganin, Evgeniya Ustinova, Hana Ajakan, Pas- cal Germain, Hugo Larochelle, Franc ¸ois Laviolette, Mario Marchand, and Victor Lempitsky. Domain-adversarial train- ing of neural networks. The Journal of Machine Learning Research, 17(1):2096–2030, 2016. 3

    2016

  9. [16]

    Imagenet-trained cnns are biased towards texture; increas- ing shape bias improves accuracy and robustness

    Robert Geirhos, Fabian Rubisch, Claudio Michaelis, Matthias Bethge, Felix A Wichmann, and Wieland Brendel. Imagenet-trained cnns are biased towards texture; increas- ing shape bias improves accuracy and robustness. In Inter- national Conference on Learning Representations (ICLR) ,

  10. [17]

    Digging into self-supervised monocular depth estimation

    Cl ´ement Godard, Oisin Mac Aodha, Michael Firman, and Gabriel J Brostow. Digging into self-supervised monocular depth estimation. In Proceedings of the IEEE/CVF Interna- tional Conference on Computer Vision (ICCV), pages 3828– 3838, 2019. 2

    2019

  11. [18]

    Unsupervised domain adaptation for trans- ferring plant classification systems to new field environ- ments, crops, and robots

    Dario Gogoll, Philipp Lottes, Jan Weyler, Nik Petrinic, and Cyrill Stachniss. Unsupervised domain adaptation for trans- ferring plant classification systems to new field environ- ments, crops, and robots. In 2020 IEEE/RSJ International Conference on Intelligent Robots and Syst...

    2020

  12. [19]

    Generative adversarial nets

    Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. Generative adversarial nets. In Advances in Neural Information Processing Systems (NeurIPS), pages 2672–2680, 2014. 1

    2014

  13. [20]

    Cycada: Cycle-consistent adversarial domain adaptation

    Judy Hoffman, Eric Tzeng, Taesung Park, Jun-Yan Zhu, Phillip Isola, Kate Saenko, Alexei Efros, and Trevor Darrell. Cycada: Cycle-consistent adversarial domain adaptation. In International conference on machine learning, pages 1989–

    1989

  14. [21]

    Grid saliency for context expla- nations of semantic segmentation

    Lukas Hoyer, Manuel Munoz, Priyanka Katiyar, Anna Khoreva, and V olker Fischer. Grid saliency for context expla- nations of semantic segmentation. In NeurIPS, pages 6462– 6473, 2019. 3

    2019

  15. [22]

    Daformer: Improving network architectures and training strategies for domain-adaptive semantic segmentation

    Lukas Hoyer, Dengxin Dai, and Luc Van Gool. 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, pages 9924–9935, 2022. 4, 6

    2022

  16. [23]

    Hrda: Context-aware high-resolution domain-adaptive semantic segmentation

    Lukas Hoyer, Dengxin Dai, and Luc Van Gool. Hrda: Context-aware high-resolution domain-adaptive semantic segmentation. In European Conference on Computer Vision, pages 372–391. Springer, 2022. 1, 3, 4, 6, 7

    2022

  17. [24]

    Domain adaptive and generalizable network architectures and train- ing strategies for semantic image segmentation

    Lukas Hoyer, Dengxin Dai, and Luc Van Gool. Domain adaptive and generalizable network architectures and train- ing strategies for semantic image segmentation. arXiv preprint arXiv:2304.13615, 2023. 1

    2023 arXiv

  18. [26]

    Progressive domain adaptation for object detection

    Hung-Kuo Hsu, Chun-Han Yao, Yi-Hsuan Tsai, Wei-Chih Hung, Hung-Yu Tseng, Maneesh Singh, and Ming-Hsuan Yang. Progressive domain adaptation for object detection. In Proceedings of the IEEE/CVF Winter Conference on Ap- plications of Computer Vision, pages 749–757, 2020. 3

    2020

  19. [27]

    Squeeze-and-excitation net- works

    Jie Hu, Li Shen, and Gang Sun. Squeeze-and-excitation net- works. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , pages 7132–7141,

  20. [28]

    Unsupervised domain adaptation for weed segmentation using greedy pseudo- labelling

    Yingchao Huang and Abdul Bais. Unsupervised domain adaptation for weed segmentation using greedy pseudo- labelling. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Work- shops, pages 2484–2494, 2024. 5, 6

    2024

  21. [29]

    Ilyas, J

    T. Ilyas, J. Lee, O. Won, Y . Jeong, and H. Kim. Overcom- ing field variability: unsupervised domain adaptation for en- hanced crop-weed recognition in diverse farmlands. Fron- tiers in Plant Science, 14:1234616, 2023. 6

    2023

  22. [30]

    Semantic segmentation with generative models: Semi-supervised learning and strong out-of-domain generalization

    Daiqing Li, Junlin Yang, Karsten Kreis, Antonio Torralba, and Sanja Fidler. Semantic segmentation with generative models: Semi-supervised learning and strong out-of-domain generalization. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVP...

    2021

  23. [31]

    Cascaded feature network for semantic seg- mentation of rgb-d images

    Di Lin, Guisong Chen, Daniel Cohen-Or, Pheng-Ann Heng, and Hui Huang. Cascaded feature network for semantic seg- mentation of rgb-d images. In Proceedings of the IEEE In- ternational Conference on Computer Vision (ICCV) , 2017. 2

    2017

  24. [32]

    Learning selective self-mutual attention for rgb-d saliency detection

    Nian Liu, Ning Zhang, and Jungong Han. Learning selective self-mutual attention for rgb-d saliency detection. In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 12056–12065, 2020. 2

    2020

  25. [33]

    Fully convolutional networks for semantic segmentation

    Jonathan Long, Evan Shelhamer, and Trevor Darrell. Fully convolutional networks for semantic segmentation. In Pro- ceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 3431–3440, 2015. 2

    2015

  26. [34]

    Learning transferable features with deep adaptation networks

    Mingsheng Long, Yue Cao, Jianmin Wang, and Michael I Jordan. Learning transferable features with deep adaptation networks. In International Conference on Machine Learn- ing, pages 97–105. PMLR, 2015. 3

    2015

  27. [35]

    Decoupled weight decay regularization

    Ilya Loshchilov and Frank Hutter. Decoupled weight decay regularization. arXiv preprint arXiv:1711.05101, 2017. 6

    2017 arXiv

  28. [36]

    In- stance adaptive self-training for unsupervised domain adap- tation

    Ke Mei, Chuang Zhu, Jiaqi Zou, and Shanghang Zhang. In- stance adaptive self-training for unsupervised domain adap- tation. In ECCV, pages 415–430, 2020. 3

    2020

  29. [37]

    Weed density es- timation using semantic segmentation

    Muhammad Hamza Asad and Abdul Bais. Weed density es- timation using semantic segmentation. In Image and Video Technology: PSIVT 2019 International Workshops, Sydney, NSW, Australia, November 18–22, 2019, Revised Selected Papers 9, pages 162–171. Springer, 2020. 1

    2019

  30. [38]

    Transferrable prototypical networks for unsupervised domain adaptation

    Yingwei Pan, Ting Yao, Yehao Li, Yu Wang, Chong-Wah Ngo, and Tao Mei. Transferrable prototypical networks for unsupervised domain adaptation. In CVPR, pages 2239– 2247, 2019. 3

    2019

  31. [39]

    Vi- sion transformers for dense prediction

    Ren ´e Ranftl, Alexey Bochkovskiy, and Vladlen Koltun. Vi- sion transformers for dense prediction. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), pages 12179–12188, 2021. 2, 4, 5

    2021

  32. [40]

    sch ¨afer, Nico M

    Manuel Schwonberg, Joshua Niemeijer, Jan-Aike Term¨ohlen, J ¨org P. sch ¨afer, Nico M. Schmidt, Hanno Gottschalk, and Tim Fingscheidt. Survey on unsuper- vised domain adaptation for semantic segmentation for visual perception in automated driving. IEEE Access , 11: 54296–54336...

    2023

  33. [41]

    Correlation alignment for unsupervised domain adaptation

    Baochen Sun, Jiashi Feng, and Kate Saenko. Correlation alignment for unsupervised domain adaptation. In Domain Adaptation in Computer Vision Applications , pages 153–

  34. [42]

    Mean teachers are better role models: Weight-averaged consistency targets improve semi-supervised deep learning results

    Antti Tarvainen and Harri Valpola. Mean teachers are better role models: Weight-averaged consistency targets improve semi-supervised deep learning results. In NeurIPS, pages 1195–1204, 2017. 3, 4, 6

    2017

  35. [43]

    Dacs: Domain adaptation via cross-domain mixed sampling

    William Tranheden, Viktor Olsson, Joao Pinto, and Lars Svensson. Dacs: Domain adaptation via cross-domain mixed sampling. In Proceedings of the Winter Conference on Ap- plications of Computer Vision (WACV) , pages 1379–1389,

  36. [44]

    Learning to adapt structured output space for semantic seg- mentation

    Yi-Hsuan Tsai, Wei-Chih Hung, Samuel Schulter, Ki- hyuk Sohn, Ming-Hsuan Yang, and Manmohan Chandraker. Learning to adapt structured output space for semantic seg- mentation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 747...

    2018

  37. [45]

    Gustavo J. Q. Vasconcelos, Thiago V . Spina, and Helio Pedrini. Low-cost domain adaptation for crop and weed seg- mentation. In Progress in Pattern Recognition, Image Anal- ysis, Computer Vision, and Applications , pages 141–150, Cham, 2021. Springer International Publishing. 1, 6

    2021

  38. [46]

    In- troduction to domain adaptation

    Hemanth Venkateswara and Sethuraman Panchanathan. In- troduction to domain adaptation. In Domain Adaptation in Computer Vision with Deep Learning, pages 3–21. Springer International Publishing, Cham, 2020. 1

    2020

  39. [47]

    Deep hashing network for unsupervised domain adaptation

    Hemanth Venkateswara, Jose Eusebio, Shayok Chakraborty, and Sethuraman Panchanathan. Deep hashing network for unsupervised domain adaptation. InProceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 5018–5027, 2017. Dataset URL: https: / / ...

    2017

  40. [48]

    Dada: Depth-aware domain adap- tation in semantic segmentation

    Tuan-Hung Vu, Himalaya Jain, Maxime Bucher, Matthieu Cord, and Patrick P ´erez. Dada: Depth-aware domain adap- tation in semantic segmentation. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), pages 7364–7373, 2019. 3

    2019

  41. [49]

    Evo- lutionary generative adversarial networks

    Chaoyue Wang, Chang Xu, Xin Yao, and Dacheng Tao. Evo- lutionary generative adversarial networks. IEEE Transac- tions on Evolutionary Computation, 23(6):921–934, 2019. 3

    2019

  42. [50]

    Classes matter: A fine-grained adversarial ap- proach to cross-domain semantic segmentation

    Haoran Wang, Tong Shen, Wei Zhang, Ling-Yu Duan, and Tao Mei. Classes matter: A fine-grained adversarial ap- proach to cross-domain semantic segmentation. InEuropean conference on computer vision , pages 642–659. Springer,

  43. [51]

    Learning common and specific features for rgb- d semantic segmentation with deconvolutional networks

    Jingdong Wang, Ziwei Wang, Dacheng Tao, Simon See, and Gang Wang. Learning common and specific features for rgb- d semantic segmentation with deconvolutional networks. In European Conference on Computer Vision (ECCV) , pages 664–679, 2016. 2

    2016

  44. [52]

    Understanding convolution for semantic segmentation

    Panqu Wang, Pengfei Chen, Ye Yuan, Ding Liu, Zehua Huang, Xiaodi Hou, and Garrison Cottrell. Understanding convolution for semantic segmentation. In Proceedings of the IEEE Winter Conference on Applications of Computer Vision (WACV), pages 1451–1460, 2018. 2

    2018

  45. [53]

    Depth-aware cnn for rgb-d segmentation

    Weiyue Wang and Ulrich Neumann. Depth-aware cnn for rgb-d segmentation. In European Conference on Computer Vision (ECCV), 2018. 2

    2018

  46. [54]

    Masked feature predic- tion for self-supervised visual pre-training

    Chen Wei, Haoqi Fan, Saining Xie, Chao-Yuan Wu, Alan Yuille, and Christoph Feichtenhofer. Masked feature predic- tion for self-supervised visual pre-training. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 14668–14678, 2022. 3

    2022

  47. [55]

    A one-stage domain adaptation network with image alignment for unsupervised nighttime semantic segmentation

    Xinyi Wu, Zhenyao Wu, Lili Ju, and Song Wang. A one-stage domain adaptation network with image alignment for unsupervised nighttime semantic segmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence (PAMI), 2021. 3

    2021

  48. [56]

    Segformer: Simple and efficient design for semantic segmentation with transform- ers

    Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M Alvarez, and Ping Luo. Segformer: Simple and efficient design for semantic segmentation with transform- ers. Advances in Neural Information Processing Systems , 34:12077–12090, 2021. 5

    2021

  49. [57]

    Unifying flow, stereo and depth estimation

    Haofei Xu, Jiaming Zhang, Jianfei Cai, Hamid Rezatofighi, Fisher Yu, Dacheng Tao, and Andreas Geiger. Unifying flow, stereo and depth estimation. IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI), 2023. 2, 3

    2023

  50. [58]

    Micdrop: Masking image and depth features via complementary dropout for domain- adaptive semantic segmentation

    Linyan Yang, Lukas Hoyer, Mark Weber, Tobias Fischer, Dengxin Dai, Laura Leal-Taix ´e, Marc Pollefeys, Daniel Cremers, and Luc Van Gool. Micdrop: Masking image and depth features via complementary dropout for domain- adaptive semantic segmentation. In Computer Vision – ECCV 20...

    2024

  51. [59]

    Y . Yang, D. Lao, G. Sundaramoorthi, and S. Soatto. Phase consistent ecological domain adaptation. In 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 9008–9017, Los Alamitos, CA, USA, 2020. IEEE Computer Society. 6

    2020

  52. [60]

    Cmx: Cross-modal fusion for rgb-x semantic segmentation with transformers.IEEE Trans- actions on Intelligent Transportation Systems, 2023

    Jing Zhang, Hongyu Liu, Kai Yang, Xiaolin Hu, Risheng Liu, and Rainer Stiefelhagen. Cmx: Cross-modal fusion for rgb-x semantic segmentation with transformers.IEEE Trans- actions on Intelligent Transportation Systems, 2023. 2

    2023

  53. [61]

    Cat- egory anchor-guided unsupervised domain adaptation for se- mantic segmentation

    Qiming Zhang, Jing Zhang, Wei Liu, and Dacheng Tao. Cat- egory anchor-guided unsupervised domain adaptation for se- mantic segmentation. In NeurIPS, pages 435–445, 2019. 3

    2019

  54. [62]

    Pattern-affinitive propagation across depth, surface normal and semantic segmentation

    Zhaoxiang Zhang, Zhen Cui, Chen Xu, Yan Yan, Nicu Sebe, and Jie Yang. Pattern-affinitive propagation across depth, surface normal and semantic segmentation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pat- tern Recognition (CVPR), 2019. 2

    2019

  55. [63]

    Unpaired image-to-image translation using cycle- consistent adversarial networks

    Jun-Yan Zhu, Taesung Park, Phillip Isola, and Alexei A Efros. Unpaired image-to-image translation using cycle- consistent adversarial networks. In Proceedings of the IEEE international conference on computer vision , pages 2223– 2232, 2017. 3

    2017

  56. [64]

    Un- supervised domain adaptation for semantic segmentation via class-balanced self-training

    Yang Zou, Zhiding Yu, BVK Kumar, and Jinsong Wang. Un- supervised domain adaptation for semantic segmentation via class-balanced self-training. In ECCV, pages 289–305, 2018. 6

    2018

  57. [2021]

    Sections 2, 3, 4, 6, 8. 3, 4, 5

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.