Pith. sign in

REVIEW 4 major objections 5 minor 93 references

SoMA: Singular Value Decomposed Minor Components Adaptation for Domain Generalizable Representation Learning

T0 review · 4 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

Pith's one-line read SoMA claims that tuning only the smallest singular components of a vision model's weights preserves its generalization while still learning task skills, and demonstrates state-of-the-art domain generalization on segmentation and detection…

desk verdict SoMA is a genuinely novel PEFT idea with consistent ablation support, but its 'minor-component tuning' is an initialization scheme rather than a structural constraint, and the empirical reporting needs more rigor. read the letter →

arxiv 2412.04077 v2 pith:67RAK5BX submitted 2024-12-05 cs.CV

classification cs.CV
keywords singularvaluedecompositiondomaingeneralizationparameter-efficientfine-tuninglow-rankadaptationsemanticsegmentationobjectdetectionvisionfoundationmodelsrepresentationlearning
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 argues that the reason parameter-efficient fine-tuning can preserve a vision foundation model's ability to generalize to unseen domains is that most of the model's world knowledge lives in the largest singular components of its weight matrices. It introduces SoMA, which decomposes each pre-trained weight matrix with SVD, keeps the principal components frozen, and initializes a LoRA-style adapter from the smallest $r$ singular components so that gradients modify only those weak directions. Applied to domain-generalized semantic segmentation and object detection, SoMA is claimed to outperform full fine-tuning and prior PEFT methods using 0.58% to 1.6% of backbone parameters, reaching 80.37 mIoU on GTAV+SYNTHIA+UrbanSyn to Cityscapes and 47.6 mAP on the hardest daytime-sunny to night-rainy detection shift. A reader should care because it offers a spectral criterion for where to spend adaptation capacity: tune what is context-specific, freeze what is general.

What carries the argument

The central object is the singular value decomposition of each frozen pre-trained weight matrix, $W = U\Sigma V^T$, split into a residual part $W_{\mathrm{res}}$ (top $R-r$ singular components) and a minor part $U_{[:,-r:]}\Sigma_{[-r:]}(V^T)_{[-r:,:]}$ that initializes the low-rank adapter. The QR-type initialization makes the adapter's span orthogonal to the principal singular directions, which the paper calls minimal interference with pre-trained representations. The singular modulation ratio $SMR_i = |u_i^T \Delta W v_i|/\sigma_i$ is the diagnostic that shows SoMA's learned update stays small on high singular values compared with LoRA. Freezing early blocks and annealing weight decay are secondary mechanisms that shift capacity toward discriminability without reopening the principal directions.

What would settle it

Track the singular modulation ratio $SMR_i = |u_i^\top \Delta W v_i|/\sigma_i$ during GTAV-to-Cityscapes training: if the SoMA update's projection onto the top 25% of singular directions grows to match LoRA's while Cityscapes mIoU still improves, then preserving minor-direction updates is not the mechanism behind the reported gains.

Watch

Extended reading notes

Core claim

SoMA's central claim is that the singular-value spectrum of pre-trained weights cleanly separates generalizable from context-specific knowledge, and that fine-tuning should therefore touch only the minor singular components. Starting from an analysis on ImageNet-1k, the paper reports that removing top principal components of DINOv2 weights causes broad class-level failures, whereas removing the smallest components mainly hurts fine-grained, texture-biased, and context-specific classes. SoMA sets $B = U_{[:,-r:]}\sqrt{\Sigma_{[-r:]}}$ and $A = \sqrt{\Sigma_{[-r:]}}(V^T)_{[-r:,:]}$, freezes $W_{\mathrm{res}} = W - BA$, and merges the learned $B'A'$ back at inference; it also freezes the early blocks of the backbone and anneals weight decay to zero. On GTAV-to-Cityscapes and related synthetic-to-real and real-to-real dense prediction benchmarks, the paper reports state-of-the-art mIoU and mAP while training 0.58%–1.6% of parameters, and shows quantitatively that the learned update minimally correlates with the top singular directions of the pre-trained weights.

Load-bearing premise

The paper's load-bearing premise is that the singular-value spectrum of pre-trained weights cleanly separates general from context-specific knowledge, and that the ImageNet-derived separation transfers to synthetic-to-real driving scenes; the authors themselves note in the supplement that whether focusing solely on the lowest spectral space is optimal remains open.

Editorial extensions

If this is right

  • Domain-generalized dense prediction can be done with 0.58%–1.6% of backbone parameters and no extra inference cost, replacing full fine-tuning for VFM-based systems.
  • Because the adapter is merged at inference, SoMA can be dropped into existing segmentation and detection pipelines without changing the forward pass.
  • The same recipe transfers across backbone families (DINOv2, EVA02, Swin, ConvNeXt, ResNet) and decode heads (Mask2Former, SemFPN, Co-DETR).
  • Performance scales with model and source-data size, so gains should widen as vision foundation models grow.
  • Subject personalization with diffusion models also benefits: SoMA preserves SDXL's cross-domain generation while learning a new subject.

Reading between the lines

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

  • The spectral-preservation criterion suggests a general principle for PEFT: allocate trainable capacity to the smallest singular components of any pre-trained weight matrix, not only vision transformers; a direct test would be applying SoMA to LLM fine-tuning and measuring retention of pre-training skills.
  • The SMR diagnostic could be turned into a stopping rule: halt training or shrink rank once the update's projection onto top singular directions exceeds a threshold, preventing gradual leakage into principal components.
  • The paper's own limitation note leaves open whether the lowest spectral space is optimal; an alternative worth testing is per-layer rank selection driven by the singular-value decay rate of each matrix.
  • Because the analysis is based on ImageNet class granularity, a stress test would be to construct synthetic-to-real shifts dominated by texture- and context-dependent classes to see whether minor-component tuning remains safe.
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

4 major / 5 minor

Summary. The paper proposes SoMA, a parameter-efficient fine-tuning method for domain generalization in dense prediction. SoMA performs SVD on pretrained weight matrices, initializes LoRA-style adapters from the r smallest singular components, freezes the residual weight matrix and early transformer blocks, and applies an annealing weight decay schedule. The method is motivated by an ImageNet-1k analysis that associates larger singular components with general, multi-class knowledge and smaller components with context-specific knowledge. Experiments cover domain-generalized semantic segmentation (GTAV/SYNTHIA/UrbanSyn to Cityscapes/BDD/Mapillary, Cityscapes to ACDC) and domain-generalized object detection (Daytime-Sunny to adverse weather), reporting state-of-the-art results with 0.58% to 1.6% trainable parameters, plus a DreamBooth-based subject personalization extension.

Significance. If the results hold, SoMA would be a practical and inexpensive PEFT recipe that improves over full fine-tuning and LoRA on multiple DG benchmarks with no inference overhead. The paper's strengths include consistent component-wise ablations in Tables 7, 11, and 12, evaluations across several backbones and decode heads, and a released implementation. The central mechanistic claim, however, is currently stronger than what the experiments establish: the minor-subspace behavior is only an initialization property, and the manuscript lacks statistical evidence to distinguish the reported gains from run-to-run variation. The paper is of interest to the PEFT and domain generalization communities, but the main claims need to be reframed or supported by additional analysis.

major comments (4)
  1. [Sec. 4.1] The statement that SoMA "selectively tunes minor singular components while keeping the residual parts frozen" is exactly true only at initialization. B and A are updated without any constraint, and after training W' = W_res + B'A' = W + (B'A' - BA_initial); the rank-r update Delta is not restricted to the span of U[:, -r:] and V[:, -r:]. Table 1 confirms that the top singular directions are modulated by training (SMR values of 0.075 and 0.097 for the 12th and 24th blocks), so the residual singular directions are not untouched. The authors themselves note in Supplemental Section F that whether focusing solely on the lowest spectral space is optimal remains open. I recommend either (a) modifying the method to enforce confinement (e.g., projecting gradients or adding a penalty) and showing that this retains performance, or (b) reframing the contribution as a spectral initialization/warm start and providing a direct comparison against random-initialized LoRA with the same residual split, including an analysis of where the trained update actually lives. As written, the mechanistic explanation in Sec. 3 is presented as the cause of the gains but is not supported by the reported experiments.
  2. [Tables 2-6 and 10] All main benchmark results are single runs without error bars or a stated number of seeds. Several headline improvements are small relative to the performance scale (e.g., Table 2, GTAV to BDD: SoMA 61.31 vs. LoRA 60.13; Table 6, mAP 54.3 vs. LoRA 53.3). Without variance estimates, it is unclear whether the reported margins over strong baselines are meaningful. Please report mean and standard deviation over at least three seeds for the main synthetic-to-real segmentation, real-to-real segmentation, and detection settings.
  3. [Tables 9, 10, and 13] The hyperparameters r, NFEB, and the initial weight decay coefficient appear to be selected based on target-domain test performance. The rank sweep in Table 9 reports Cityscapes/BDD/Mapillary mIoU, and the NFEB sweep in Table 13 reports Cityscapes mIoU; no separate validation set or source-domain selection criterion is described. If these choices were made using the target test domains, the DG evaluation is compromised because the target distribution was accessed during model selection. Please clarify the validation protocol and, if test-based selection was used, add a proper source-domain validation split.
  4. [Sec. 3] The "generalizable components" analysis is performed on ImageNet-1k classification with DINOv2-large, but the method is evaluated on synthetic-to-real urban dense prediction. The paper does not verify that the semantic roles of singular components (general vs. context-specific) transfer to driving-scene classes; the failure classes listed in Fig. 2 (e.g., cornet, bikini, missile, dock) are not driving classes. This is a load-bearing premise for the central claim. Please either provide a similar SVD-truncation analysis on a driving/scene dataset (e.g., GTAV or Cityscapes features) or temper the claim that the ImageNet analysis explains the DGSS results.
minor comments (5)
  1. [Abstract and Sec. 1] There is a typo in the phrase "Singular Value Dec omposed" in the Section 1 heading; also, the notation "EV A02" appears with inconsistent spacing throughout the paper. Please standardize.
  2. [Eq. (1)] The text refers to "the underlined low-rank parameters," but the equation is not typeset with underlining in the submitted version; please make the trainable parameters explicit in the notation.
  3. [Sec. 5.2, Table 2] The marker definitions for "*", "•", "†", and "‡" are dense and easy to miss; in particular, "◦" is used in tables but not explicitly defined in the marker list. Please add a clear legend directly below each main table.
  4. [Table 4] The row for VLTSeg reports dashes for the Night, Snow, Fog, and Rain columns; please indicate in the caption why these values are omitted.
  5. [Supplemental Sec. A.2] The sentence "we linearize both the patch-level convolution and its weights" is a bit ambiguous; it would be clearer to state explicitly that the convolution is unrolled into a matrix multiplication and SoMA is applied to that matrix.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: SoMA's spectral initialization is not fitted to target benchmarks, and the DG results are measured on held-out domains.

full rationale

The derivation chain is self-contained. SoMA initializes its low-rank adapter from the smallest-r singular components of the frozen pre-trained weight matrix W via Eqs. (4)-(5), sets Wres = W - BA, and then trains B and A. The final effective update is B'A' - BA, and performance is reported on unseen target domains (Cityscapes, BDD100K, Mapillary, ACDC, and adverse-weather detection); none of those target-domain numbers are used to fit the rank r, the number of frozen early blocks, or the annealing weight-decay schedule. The Sec. 3 ImageNet SVD analysis is an external empirical observation about which singular components, when removed, hurt which classes; it is not an input to the training objective, and the paper does not define 'generalizable' to mean 'high singular value' in a way that makes the later design step tautological. The SMR metric in Eq. (7) is a post-training diagnostic: at initialization the update is zero, but the reported values are measured after training and merely quantify the correlation of the learned update with W's singular directions. That the initialization is confined to minor directions is true by construction, but the paper does not relabel that construction as a benchmark prediction. The only substantive caveat is that 'keeping the residual parts frozen' is literal at the parameter level yet does not mathematically guarantee that the trained update B'A' - BA remains in the minor singular subspace; the paper's own limitation statement in Supplemental Sec. F acknowledges that whether tuning only the lowest spectral space is optimal remains open. That is a correctness or mechanism-support concern, not a circularity, because the central empirical claims are external benchmark comparisons rather than consequences of the paper's definitions.

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

The central mechanism rests on an empirical map between singular-value magnitude and semantic generality. Rank r, block-freeze count, and the initial weight decay coefficient are chosen by the authors, while the spectral separation, block-freeze principle, and early-learning weight-decay behavior are domain assumptions borrowed from analysis or prior literature. No new physical or conceptual entity is introduced.

free parameters (3)
  • rank r = 16
    Number of minor singular components tuned per linear layer; selected via GTAV to Cityscapes ablation (Table 9), with 8 used for DINOv2-G and 24 for ResNet101 detection.
  • NFEB (number of frozen early blocks) = 8
    Chosen from GTAV to Cityscapes ablation over {0, 4, 8, 12, 16} (Table 13); 12 for DINOv2-G and 21 for ResNet101.
  • initial weight decay coefficient = 5e-2 (3e-2 for EVA02)
    Starting value of the annealing weight decay schedule; a hand-chosen hyperparameter reported in Table 10.
assumptions (4)
  • standard math Eckart-Young-Mirsky theorem guarantees that the optimal rank-r approximation of a matrix is given by its top r singular components.
    Invoked in Sec. 3 to justify analyzing the weight matrix through truncation of singular components.
  • domain assumption Singular vectors associated with higher singular values extract general features spanning multiple classes, while minor singular components capture context-specific features.
    Empirical claim from ImageNet-1k truncation experiments (Fig. 2); it is the premise that makes tuning minor components the correct preservation strategy.
  • domain assumption Freezing early blocks of vision foundation models preserves domain-invariant semantics and improves generalization.
    Supported by PCA and rare-class IoU in Fig. 3 and by the NFEB ablation, but assumed as a general principle in Sec. 4.2.
  • domain assumption Weight decay primarily shapes early learning dynamics, so annealing it to zero improves discriminability without harming generalization.
    Based on cited prior work [26] and adopted in Sec. 4.3 to justify the annealing weight decay scheme.

how reviews work

0 comments
Cite this review

Pith. "Pith review of SoMA: Singular Value Decomposed Minor Components Adaptation for Domain Generalizable Representation Learning." pith.science (2026). https://pith.science/paper/67RAK5BX

@misc{pith2026241204077,
  author       = {Pith},
  title        = {Pith review of: SoMA: Singular Value Decomposed Minor Components Adaptation for Domain Generalizable Representation Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/67RAK5BX}},
  note         = {Machine review of arXiv:2412.04077}
}
read the original abstract

Domain generalization (DG) aims to adapt a model using one or multiple source domains to ensure robust performance in unseen target domains. Recently, Parameter-Efficient Fine-Tuning (PEFT) of foundation models has shown promising results in the context of DG problem. Nevertheless, existing PEFT methods still struggle to strike a balance between preserving generalizable components of the pre-trained model and learning task-specific features. To gain insights into the distribution of generalizable components, we begin by analyzing the pre-trained weights through the lens of singular value decomposition. Building on these insights, we introduce Singular Value Decomposed Minor Components Adaptation (SoMA), an approach that selectively tunes minor singular components while keeping the residual parts frozen. SoMA effectively retains the generalization ability of the pre-trained model while efficiently acquiring task-specific skills. Moreover, we freeze domain-generalizable blocks and employ an annealing weight decay strategy, thereby achieving an optimal balance in the delicate trade-off between generalizability and discriminability. SoMA attains state-of-the-art results on multiple benchmarks that span both domain generalized semantic segmentation to domain generalized object detection. In addition, our methods introduce no additional inference overhead or regularization loss, maintain compatibility with any backbone or head, and are designed to be versatile, allowing easy integration into a wide range of tasks.

Figures

Figures reproduced from arXiv: 2412.04077 by the authors.

Figure 1
Figure 1. Overview of SoMA framework. Left. SoMA achieves state-of-the-art results across diverse tasks, ranging from domain￾generalized semantic segmentation (DGSS) to object detection (DGOD), and performs well in both synthetic-to-real and real-to-real sce￾narios. Middle. Our method, trained solely on synthetic datasets, demonstrates strong generalization capabilities in complex real-world scenes. Right. We present SoMA, a … view at source ↗
Figure 2
Figure 2. Distribution of generalizable components. Top. Num￾ber of classes exhibiting specific accuracy drops after applying SVD to DINOv2-large weights and reconstructing by truncating the smallest r singular components. Bottom. Distinct roles of sin￾gular components across levels. Numbers in parentheses represent the count of classes with an accuracy drop ratio exceeding 50%. The average WordNet hierarchy depth of these cl… view at source ↗
Figure 3
Figure 3. The inherent generalization capabilities of the early blocks of VFM. Top. We apply PCA on the extracted intermediate features (8th block of DINOv2-Large) and visualize the top three leading components. Bottom. Class-wise IoU comparison for rare classes [32] under the GTAV→Cityscapes setting, focusing on the impact of freezing the first eight blocks. “NFEB” stands for the Number of Frozen Early Blocks. A darker shade… view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: DGOD qualitative results. Methods Params.∗ DGSS Avg. DGOD Avg. Full fine-tuning (baseline) 304.2M 64.4 51.0 \delimiter "4478478 + Freezing early blocks 201.6M 65.0 (↑ 0.6) 51.4 (↑ 0.4) \delimiter "4478478 + Tuning principal components 4.9M 66.1 (↑ 1.1) 53.0 (↑ 1.6) \de…
Figure 6
Figure 6. Figure 6: Segmentation results of SoMA on the Cityscapes. The model is trained on GTAV with DINOv2-L backbone. [PITH_FULL_IMAGE:figures/full_fig_p014_6.png]
Figure 7
Figure 7. Figure 7: Segmentation results of SoMA on the BDD100k. The model is trained on GTAV with DINOv2-L backbone. [PITH_FULL_IMAGE:figures/full_fig_p014_7.png]
Figure 8
Figure 8. Figure 8: Segmentation results of SoMA on the Mapillary. The model is trained on GTAV with DINOv2-L backbone. [PITH_FULL_IMAGE:figures/full_fig_p015_8.png]
Figure 9
Figure 9. Figure 9: Detection results of SoMA on the adverse scene. The model is trained on Daytime-Sunny with DINOv2-L backbone. [PITH_FULL_IMAGE:figures/full_fig_p015_9.png]
Figure 10
Figure 10. Figure 10: Multiple subject-consistent synthesis results with prompts describing various domains. SoMA effectively preserves SDXL’s ability to generate images across diverse domains while learning new visual concepts. As a result, simply using prompts from multiple domains allow…
Figure 11
Figure 11. Figure 11: Qualitative comparison to DreamBooth [70] with prior preservation (p.p.) loss. Usunier, Alexander Kirillov, and Sergey Zagoruyko. End-to￾end object detection with transformers. In ECCV, 2020. 1 [8] Mathilde Caron, Hugo Touvron, Ishan Misra, Herve J ´ egou, ´ Julien Ma…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

93 extracted references · 69 canonical work pages

  1. [1]

    Style blind domain generalized semantic segmentation via covariance alignment and semantic consis- tence contrastive learning

    Woo-Jin Ahn, Geun-Yeong Yang, Hyun-Duck Choi, and Myo-Taeg Lim. Style blind domain generalized semantic segmentation via covariance alignment and semantic consis- tence contrastive learning. In CVPR, 2024. 2, 3, 13

  2. [2]

    Foundational models defining a new era in vision: A survey and outlook

    Muhammad Awais, Muzammal Naseer, Salman Khan, Rao Muhammad Anwer, Hisham Cholakkal, Mubarak Shah, Ming-Hsuan Yang, and Fahad Shahbaz Khan. Foundational models defining a new era in vision: A survey and outlook. arXiv:2307.13721, 2023. 2

  3. [3]

    Collaborating foundation models for domain generalized semantic segmentation

    Yasser Benigmim, Subhankar Roy, Slim Essid, Vicky Kalo- geiton, and St ´ephane Lathuili`ere. Collaborating foundation models for domain generalized semantic segmentation. In CVPR, 2024. 3, 6, 13

  4. [4]

    Learning frequency- adapted vision foundation model for domain generalized se- mantic segmentation

    Qi Bi, Jingjun Yi, Hao Zheng, Haolan Zhan, Yawen Huang, Wei Ji, Yuexiang Li, and Yefeng Zheng. Learning frequency- adapted vision foundation model for domain generalized se- mantic segmentation. In NeurIPS, 2024. 3, 6, 7, 13

  5. [5]

    Learning content- enhanced mask transformer for domain generalized urban- scene segmentation

    Qi Bi, Shaodi You, and Theo Gevers. Learning content- enhanced mask transformer for domain generalized urban- scene segmentation. In AAAI, 2024. 2, 7, 13

  6. [6]

    Lora learns less and forgets less

    Dan Biderman, Jose Gonzalez Ortiz, Jacob Portes, Mansheej Paul, Philip Greengard, Connor Jennings, Daniel King, Sam Havens, Vitaliy Chiley, Jonathan Frankle, et al. Lora learns less and forgets less. TMLR, 2024. 2

  7. [7]

    Segmentation results of SoMA on the Cityscapes

    Nicolas Carion, Francisco Massa, Gabriel Synnaeve, Nicolas 13 Figure 6. Segmentation results of SoMA on the Cityscapes. The model is trained on GTA V with DINOv2-L backbone. Figure 7. Segmentation results of SoMA on the BDD100k. The model is trained on GTA V with DINOv2-L backbone. 14 Figure 8. Segmentation results of SoMA on the Mapillary. The model is t...

  8. [8]

    Emerg- ing properties in self-supervised vision transformers

    Mathilde Caron, Hugo Touvron, Ishan Misra, Herv ´e J´egou, Julien Mairal, Piotr Bojanowski, and Armand Joulin. Emerg- ing properties in self-supervised vision transformers. In ICCV, 2021. 3

Show all 93 references
  1. [9]

    Unified domain generalization and adaptation for multi-view 3d object detection

    Gyusam Chang, Jiwon Lee, Donghyun Kim, Jinkyu Kim, Dongwook Lee, Daehyun Ji, Sujin Jang, and Sangpil Kim. Unified domain generalization and adaptation for multi-view 3d object detection. In NeurIPS, 2024. 2

  2. [10]

    Pasta: Proportional amplitude spectrum training augmentation for syn-to-real domain gen- eralization

    Prithvijit Chattopadhyay, Kartik Sarangmath, Vivek Vi- jaykumar, and Judy Hoffman. Pasta: Proportional amplitude spectrum training augmentation for syn-to-real domain gen- eralization. In ICCV, 2023. 2, 3, 13 16

  3. [11]

    MMDe- tection: Open mmlab detection toolbox and benchmark

    Kai Chen, Jiaqi Wang, Jiangmiao Pang, Yuhang Cao, Yu Xiong, Xiaoxiao Li, Shuyang Sun, Wansen Feng, Ziwei Liu, Jiarui Xu, Zheng Zhang, Dazhi Cheng, Chenchen Zhu, Tianheng Cheng, Qijie Zhao, Buyu Li, Xin Lu, Rui Zhu, Yue Wu, Jifeng Dai, Jingdong Wang, Jianping Shi, Wanli Ouyang,...

  4. [12]

    Adaptformer: Adapt- ing vision transformers for scalable visual recognition

    Shoufa Chen, Chongjian Ge, Zhan Tong, Jiangliu Wang, Yibing Song, Jue Wang, and Ping Luo. Adaptformer: Adapt- ing vision transformers for scalable visual recognition. In NeurIPS, 2022. 2, 3, 6, 7, 13

  5. [13]

    Masked-attention mask transformer for universal image segmentation

    Bowen Cheng, Ishan Misra, Alexander G Schwing, Alexan- der Kirillov, and Rohit Girdhar. Masked-attention mask transformer for universal image segmentation. In CVPR,

  6. [14]

    Robustnet: Improving domain generalization in urban-scene segmentation via in- stance selective whitening

    Sungha Choi, Sanghun Jung, Huiwon Yun, Joanne T Kim, Seungryong Kim, and Jaegul Choo. Robustnet: Improving domain generalization in urban-scene segmentation via in- stance selective whitening. In CVPR, 2021. 2, 3, 5, 13

  7. [15]

    MMSegmentation: Openmmlab semantic segmentation toolbox and benchmark

    MMSegmentation Contributors. MMSegmentation: Openmmlab semantic segmentation toolbox and benchmark. https : / / github . com / open - mmlab/mmsegmentation, 2020. 9

  8. [16]

    The cityscapes dataset for semantic urban scene understanding

    Marius Cordts, Mohamed Omran, Sebastian Ramos, Timo Rehfeld, Markus Enzweiler, Rodrigo Benenson, Uwe Franke, Stefan Roth, and Bernt Schiele. The cityscapes dataset for semantic urban scene understanding. In CVPR,

  9. [17]

    Improving single domain-generalized object detection: A focus on diversification and alignment

    Muhammad Sohail Danish, Muhammad Haris Khan, Muhammad Akhtar Munir, M Saquib Sarfraz, and Mohsen Ali. Improving single domain-generalized object detection: A focus on diversification and alignment. In CVPR, 2024. 3, 7, 9

  10. [18]

    Imagenet: A large-scale hierarchical image database

    Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. Imagenet: A large-scale hierarchical image database. In CVPR, 2009. 2, 4, 11

  11. [19]

    Hgformer: Hierarchical grouping transformer for domain generalized semantic segmentation

    Jian Ding, Nan Xue, Gui-Song Xia, Bernt Schiele, and Dengxin Dai. Hgformer: Hierarchical grouping transformer for domain generalized semantic segmentation. In CVPR,

  12. [20]

    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, Jakob Uszkoreit, and Neil Houlsby. An image is worth 16x16 words: Transformers for image recognition ...

  13. [21]

    The approximation of one ma- trix by another of lower rank

    Carl Eckart and Gale Young. The approximation of one ma- trix by another of lower rank. Psychometrika, 1936. 3

  14. [22]

    A simple recipe for language- guided domain generalized segmentation

    Mohammad Fahes, Tuan-Hung Vu, Andrei Bursuc, Patrick P´erez, and Raoul de Charette. A simple recipe for language- guided domain generalized segmentation. In CVPR, 2024. 3, 13

  15. [23]

    Towards robust ob- ject detection invariant to real-world domain shifts

    Qi Fan, Mattia Segu, Yu-Wing Tai, Fisher Yu, Chi-Keung Tang, Bernt Schiele, and Dengxin Dai. Towards robust ob- ject detection invariant to real-world domain shifts. InICLR,

  16. [24]

    Eva-02: A visual representation for neon genesis

    Yuxin Fang, Quan Sun, Xinggang Wang, Tiejun Huang, Xin- long Wang, and Yue Cao. Eva-02: A visual representation for neon genesis. IVC, 2024. 3, 6, 11

  17. [25]

    Implicit style-content separation using b-lora

    Yarden Frenkel, Yael Vinker, Ariel Shamir, and Daniel Cohen-Or. Implicit style-content separation using b-lora. In ECCV, 2024. 9

  18. [26]

    Time matters in regularizing deep networks: Weight decay and data augmentation affect early learning dynamics, matter little near convergence

    Aditya Sharad Golatkar, Alessandro Achille, and Stefano Soatto. Time matters in regularizing deep networks: Weight decay and data augmentation affect early learning dynamics, matter little near convergence. In NeurIPS, 2019. 6

  19. [27]

    All for one, and one for all: Urbansyn dataset, the third musketeer of synthetic driv- ing scenes

    Jose L G ´omez, Manuel Silva, Antonio Seoane, Agn `es Borr´as, Mario Noriega, Germ ´an Ros, Jose A Iglesias- Guitian, and Antonio M L ´opez. All for one, and one for all: Urbansyn dataset, the third musketeer of synthetic driv- ing scenes. arXiv:2312.12176, 2023. 1, 6

  20. [28]

    Towards a unified view of parameter-efficient transfer learning

    Junxian He, Chunting Zhou, Xuezhe Ma, Taylor Berg- Kirkpatrick, and Graham Neubig. Towards a unified view of parameter-efficient transfer learning. In ICLR, 2022. 2

  21. [29]

    Delving deep into rectifiers: Surpassing human-level perfor- mance on imagenet classification

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Delving deep into rectifiers: Surpassing human-level perfor- mance on imagenet classification. In ICCV, 2015. 3

  22. [30]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In CVPR,

  23. [31]

    Masked autoencoders are scalable vision learners

    Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Doll´ar, and Ross Girshick. Masked autoencoders are scalable vision learners. In CVPR, 2022. 3, 11

  24. [32]

    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 CVPR, 2022. 5

  25. [33]

    LoRA: Low-rank adaptation of large language models

    Edward J Hu, yelong shen, Phillip Wallis, Zeyuan Allen- Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. LoRA: Low-rank adaptation of large language models. In ICLR, 2022. 1, 2, 3, 5, 6, 7, 11, 13

  26. [34]

    Learn to preserve and diversify: Parameter-efficient group with orthogonal regularization for domain generalization

    Jiajun Hu, Jian Zhang, Lei Qi, Yinghuan Shi, and Yang Gao. Learn to preserve and diversify: Parameter-efficient group with orthogonal regularization for domain generalization. In ECCV, 2024. 2, 11

  27. [35]

    Style pro- jected clustering for domain generalized semantic segmenta- tion

    Wei Huang, Chang Chen, Yong Li, Jiacheng Li, Cheng Li, Fenglong Song, Youliang Yan, and Zhiwei Xiong. Style pro- jected clustering for domain generalized semantic segmenta- tion. In CVPR, 2023. 3, 13

  28. [36]

    Vltseg: Sim- ple transfer of clip-based vision-language representations for domain generalized semantic segmentation

    Christoph H ¨ummer, Manuel Schwonberg, Liangwei Zhong, Hu Cao, Alois Knoll, and Hanno Gottschalk. Vltseg: Sim- ple transfer of clip-based vision-language representations for domain generalized semantic segmentation. In ACCV, 2024. 3, 6, 7, 13

  29. [37]

    Vi- sual prompt tuning

    Menglin Jia, Luming Tang, Bor-Chun Chen, Claire Cardie, Serge Belongie, Bharath Hariharan, and Ser-Nam Lim. Vi- sual prompt tuning. In ECCV, 2022. 2, 3, 6, 13

  30. [38]

    Dgin- style: Domain-generalizable semantic segmentation with image diffusion models and stylized semantic control

    Yuru Jia, Lukas Hoyer, Shengyu Huang, Tianfu Wang, Luc Van Gool, Konrad Schindler, and Anton Obukhov. Dgin- style: Domain-generalizable semantic segmentation with image diffusion models and stylized semantic control. In ECCV, 2024. 3, 13

  31. [39]

    Domain generalization via balancing training difficulty and model capability

    Xueying Jiang, Jiaxing Huang, Sheng Jin, and Shijian Lu. Domain generalization via balancing training difficulty and model capability. In ICCV, 2023. 3, 13 17

  32. [40]

    Order-preserving consistency regularization for do- main adaptation and generalization

    Mengmeng Jing, Xiantong Zhen, Jingjing Li, and Cees GM Snoek. Order-preserving consistency regularization for do- main adaptation and generalization. In ICCV, 2023. 3

  33. [41]

    Increasing the ro- bustness of semantic segmentation models with painting-by- numbers

    Christoph Kamann and Carsten Rother. Increasing the ro- bustness of semantic segmentation models with painting-by- numbers. In ECCV, 2020. 3

  34. [42]

    Pin the memory: Learning to generalize semantic segmentation

    Jin Kim, Jiyoung Lee, Jungin Park, Dongbo Min, and Kwanghoon Sohn. Pin the memory: Learning to generalize semantic segmentation. In CVPR, 2022. 3, 13

  35. [43]

    Tex- ture learning domain randomization for domain generalized segmentation

    Sunghwan Kim, Dae-hwan Kim, and Hoseong Kim. Tex- ture learning domain randomization for domain generalized segmentation. In ICCV, 2023. 3, 5, 13

  36. [44]

    Panoptic feature pyramid networks

    Alexander Kirillov, Ross Girshick, Kaiming He, and Piotr Doll´ar. Panoptic feature pyramid networks. In CVPR, 2019. 11

  37. [45]

    Wildnet: Learning domain generalized semantic seg- mentation from the wild

    Suhyeon Lee, Hongje Seong, Seongwon Lee, and Euntai Kim. Wildnet: Learning domain generalized semantic seg- mentation from the wild. In CVPR, 2022. 3, 5, 13

  38. [46]

    Fifo: Learn- ing fog-invariant features for foggy scene segmentation

    Sohyun Lee, Taeyoung Son, and Suha Kwak. Fifo: Learn- ing fog-invariant features for foggy scene segmentation. In CVPR, 2022. 3

  39. [47]

    Object-aware domain generalization for object detection

    Wooju Lee, Dasol Hong, Hyungtae Lim, and Hyun Myung. Object-aware domain generalization for object detection. In AAAI, 2024. 3, 7

  40. [48]

    Surgical fine- tuning improves adaptation to distribution shifts

    Yoonho Lee, Annie S Chen, Fahim Tajwar, Ananya Kumar, Huaxiu Yao, Percy Liang, and Chelsea Finn. Surgical fine- tuning improves adaptation to distribution shifts. In ICLR,

  41. [49]

    The power of scale for parameter-efficient prompt tuning

    Brian Lester, Rami Al-Rfou, and Noah Constant. The power of scale for parameter-efficient prompt tuning. In EMNLP,

  42. [50]

    Prompt-driven dynamic object-centric learning for single do- main generalization

    Deng Li, Aming Wu, Yaowei Wang, and Yahong Han. Prompt-driven dynamic object-centric learning for single do- main generalization. In CVPR, 2024. 3, 7

  43. [51]

    Scaling & shifting your features: A new baseline for efficient model tuning

    Dongze Lian, Daquan Zhou, Jiashi Feng, and Xinchao Wang. Scaling & shifting your features: A new baseline for efficient model tuning. In NeurIPS, 2022. 3, 6, 13

  44. [52]

    DoRA: Weight-decomposed low-rank adaptation

    Shih-Yang Liu, Chien-Yi Wang, Hongxu Yin, Pavlo Molchanov, Yu-Chiang Frank Wang, Kwang-Ting Cheng, and Min-Hung Chen. DoRA: Weight-decomposed low-rank adaptation. In ICML, 2024. 3, 6, 7, 13

  45. [53]

    Unbiased faster r-cnn for single- source domain generalized object detection

    Yajing Liu, Shijun Zhou, Xiyao Liu, Chunhui Hao, Baojie Fan, and Jiandong Tian. Unbiased faster r-cnn for single- source domain generalized object detection. In CVPR, 2024. 3, 7

  46. [54]

    Swin transformer: Hierarchical vision transformer using shifted windows

    Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, and Baining Guo. Swin transformer: Hierarchical vision transformer using shifted windows. In ICCV, 2021. 11

  47. [55]

    Pissa: Prin- cipal singular values and singular vectors adaptation of large language models

    Fanxu Meng, Zhaohui Wang, and Muhan Zhang. Pissa: Prin- cipal singular values and singular vectors adaptation of large language models. In NeurIPS, 2024. 3, 8, 9, 11

  48. [56]

    The mapillary vistas dataset for semantic understanding of street scenes

    Gerhard Neuhold, Tobias Ollmann, Samuel Rota Bulo, and Peter Kontschieder. The mapillary vistas dataset for semantic understanding of street scenes. In ICCV, 2017. 6

  49. [57]

    Generalization by adaptation: Diffusion-based domain extension for domain-generalized semantic segmentation

    Joshua Niemeijer, Manuel Schwonberg, Jan-Aike Term¨ohlen, Nico M Schmidt, and Tim Fingscheidt. Generalization by adaptation: Diffusion-based domain extension for domain-generalized semantic segmentation. In WACV, 2024. 3, 13

  50. [58]

    Dinov2: Learning robust visual features without supervision

    Maxime Oquab, Timoth ´ee Darcet, Th ´eo Moutakanni, Huy V o, Marc Szafraniec, Vasil Khalidov, Pierre Fernandez, Daniel Haziza, Francisco Massa, Alaaeldin El-Nouby, et al. Dinov2: Learning robust visual features without supervision. TMLR, 2023. 3, 4, 5, 6, 7, 9, 11

  51. [59]

    Textual query-driven mask transformer for domain generalized segmentation

    Byeonghyun Pak, Byeongju Woo, Sunghwan Kim, Dae- hwan Kim, and Hoseong Kim. Textual query-driven mask transformer for domain generalized segmentation. In ECCV,

  52. [60]

    Two at once: Enhancing learning and generalization capacities via ibn-net

    Xingang Pan, Ping Luo, Jianping Shi, and Xiaoou Tang. Two at once: Enhancing learning and generalization capacities via ibn-net. In ECCV, 2018. 3, 13

  53. [61]

    Exploring reliable matching with phase enhancement for night-time semantic segmentation

    Yuwen Pan, Rui Sun, Naisong Luo, Tianzhu Zhang, and Yongdong Zhang. Exploring reliable matching with phase enhancement for night-time semantic segmentation. In ECCV, 2024. 3

  54. [62]

    Global and local texture randomization for synthetic-to-real semantic segmentation

    Duo Peng, Yinjie Lei, Lingqiao Liu, Pingping Zhang, and Jun Liu. Global and local texture randomization for synthetic-to-real semantic segmentation. TIP, 2021. 2, 3, 13

  55. [63]

    Semantic-aware domain generalized segmentation

    Duo Peng, Yinjie Lei, Munawar Hayat, Yulan Guo, and Wen Li. Semantic-aware domain generalized segmentation. In CVPR, 2022. 2, 3, 13

  56. [64]

    SDXL: Improving latent diffusion mod- els for high-resolution image synthesis

    Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas M ¨uller, Joe Penna, and Robin Rombach. SDXL: Improving latent diffusion mod- els for high-resolution image synthesis. In ICLR, 2024. 3, 8, 9

  57. [65]

    Learn- ing transferable visual models from natural language super- vision

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. Learn- ing transferable visual models from natural language super- vision. In ICML, 2021. 3

  58. [66]

    Faster r-cnn: Towards real-time object detection with region proposal networks

    Shaoqing Ren, Kaiming He, Ross Girshick, and Jian Sun. Faster r-cnn: Towards real-time object detection with region proposal networks. In NeurIPS, 2015. 1, 7

  59. [67]

    Playing for data: Ground truth from computer games

    Stephan R Richter, Vibhav Vineet, Stefan Roth, and Vladlen Koltun. Playing for data: Ground truth from computer games. In ECCV, 2016. 1, 6

  60. [68]

    High-resolution image syn- thesis with latent diffusion models

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj¨orn Ommer. High-resolution image syn- thesis with latent diffusion models. In CVPR, 2022. 3

  61. [69]

    German Ros, Laura Sellart, Joanna Materzynska, David Vazquez, and Antonio M. Lopez. The synthia dataset: A large collection of synthetic images for semantic segmenta- tion of urban scenes. In CVPR, 2016. 1, 6

  62. [70]

    Dreambooth: Fine tuning text-to-image diffusion models for subject-driven generation

    Nataniel Ruiz, Yuanzhen Li, Varun Jampani, Yael Pritch, Michael Rubinstein, and Kfir Aberman. Dreambooth: Fine tuning text-to-image diffusion models for subject-driven generation. In CVPR, 2023. 8, 9, 16

  63. [71]

    Acdc: The adverse conditions dataset with correspondences for se- mantic driving scene understanding

    Christos Sakaridis, Dengxin Dai, and Luc Van Gool. Acdc: The adverse conditions dataset with correspondences for se- mantic driving scene understanding. In ICCV, 2021. 6, 7, 13 18

  64. [72]

    Mobilenetv2: Inverted residuals and linear bottlenecks

    Mark Sandler, Andrew Howard, Menglong Zhu, Andrey Zh- moginov, and Liang-Chieh Chen. Mobilenetv2: Inverted residuals and linear bottlenecks. In CVPR, 2018. 2

  65. [73]

    Diga: Distil to generalize and then adapt for domain adaptive semantic segmentation

    Fengyi Shen, Akhil Gurram, Ziyuan Liu, He Wang, and Alois Knoll. Diga: Distil to generalize and then adapt for domain adaptive semantic segmentation. In CVPR, 2023. 3, 13

  66. [74]

    Crossnorm and selfnorm for general- ization under distribution shifts

    Zhiqiang Tang, Yunhe Gao, Yi Zhu, Zhi Zhang, Mu Li, and Dimitris N Metaxas. Crossnorm and selfnorm for general- ization under distribution shifts. In ICCV, 2021. 3

  67. [75]

    Mrfp: Learning generalizable semantic segmentation from sim-2-real with multi-resolution feature perturbation

    Sumanth Udupa, Prajwal Gurunath, Aniruddh Sikdar, and Suresh Sundaram. Mrfp: Learning generalizable semantic segmentation from sim-2-real with multi-resolution feature perturbation. In CVPR, 2024. 2, 3, 13

  68. [76]

    Attention is all you need

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszko- reit, Llion Jones, Aidan N Gomez, Ł ukasz Kaiser, and Illia Polosukhin. Attention is all you need. In NeurIPS, 2017. 3

  69. [77]

    Clip the gap: A single domain generalization approach for object detection

    Vidit Vidit, Martin Engilberge, and Mathieu Salzmann. Clip the gap: A single domain generalization approach for object detection. In CVPR, 2023. 3, 7

  70. [78]

    Diffusers: State-of-the-art diffu- sion models

    Patrick von Platen, Suraj Patil, Anton Lozhkov, Pedro Cuenca, Nathan Lambert, Kashif Rasul, Mishig Davaadorj, Dhruv Nair, Sayak Paul, William Berman, Yiyi Xu, Steven Liu, and Thomas Wolf. Diffusers: State-of-the-art diffu- sion models. https://github.com/huggingface/ diffusers...

  71. [79]

    Disentangle then parse: Night-time se- mantic segmentation with illumination disentanglement

    Zhixiang Wei, Lin Chen, Tao Tu, Pengyang Ling, Huaian Chen, and Yi Jin. Disentangle then parse: Night-time se- mantic segmentation with illumination disentanglement. In ICCV, 2023. 3

  72. [80]

    Stronger fewer & superior: Harnessing vision foundation models for domain generalized semantic segmentation

    Zhixiang Wei, Lin Chen, Yi Jin, Xiaoxiao Ma, Tianle Liu, Pengyang Ling, Ben Wang, Huaian Chen, and Jinjin Zheng. Stronger fewer & superior: Harnessing vision foundation models for domain generalized semantic segmentation. In CVPR, 2024. 2, 3, 6, 7, 9, 11, 13

  73. [81]

    Con- vnext v2: Co-designing and scaling convnets with masked autoencoders

    Sanghyun Woo, Shoubhik Debnath, Ronghang Hu, Xinlei Chen, Zhuang Liu, In So Kweon, and Saining Xie. Con- vnext v2: Co-designing and scaling convnets with masked autoencoders. In CVPR, 2023. 11

  74. [82]

    Single-domain generalized object detection in urban scene via cyclic-disentangled self- distillation

    Aming Wu and Cheng Deng. Single-domain generalized object detection in urban scene via cyclic-disentangled self- distillation. In CVPR, 2022. 3, 6, 7

  75. [83]

    Siamdoge: Domain generalizable semantic segmen- tation using siamese network

    Zhenyao Wu, Xinyi Wu, Xiaoping Zhang, Lili Ju, and Song Wang. Siamdoge: Domain generalizable semantic segmen- tation using siamese network. In ECCV, 2022. 3, 13

  76. [84]

    Dirl: Domain-invariant representation learning for gen- eralizable semantic segmentation

    Qi Xu, Liang Yao, Zhengkai Jiang, Guannan Jiang, Wenqing Chu, Wenhui Han, Wei Zhang, Chengjie Wang, and Ying Tai. Dirl: Domain-invariant representation learning for gen- eralizable semantic segmentation. In AAAI, 2022. 3, 13

  77. [85]

    Generalized seman- tic segmentation by self-supervised source domain projec- tion and multi-level contrastive learning

    Liwei Yang, Xiang Gu, and Jian Sun. Generalized seman- tic segmentation by self-supervised source domain projec- tion and multi-level contrastive learning. In AAAI, 2023. 3

  78. [86]

    Learning spectral- decomposited tokens for domain generalized semantic seg- mentation

    Jingjun Yi, Qi Bi, Hao Zheng, Haolan Zhan, Wei Ji, Yawen Huang, Yuexiang Li, and Yefeng Zheng. Learning spectral- decomposited tokens for domain generalized semantic seg- mentation. In ACMMM, 2024. 3, 6, 11, 13

  79. [87]

    Bdd100k: A diverse driving dataset for heterogeneous multitask learning

    Fisher Yu, Haofeng Chen, Xin Wang, Wenqi Xian, Yingying Chen, Fangchen Liu, Vashisht Madhavan, and Trevor Dar- rell. Bdd100k: A diverse driving dataset for heterogeneous multitask learning. In CVPR, 2020. 6

  80. [88]

    Domain-specific block se- lection and paired-view pseudo-labeling for online test-time adaptation

    Yeonguk Yu, Sungho Shin, Seunghyeok Back, Mihwan Ko, Sangjun Noh, and Kyoobin Lee. Domain-specific block se- lection and paired-view pseudo-labeling for online test-time adaptation. In CVPR, 2024. 5

  81. [89]

    Domain randomization and pyramid consistency: Simulation-to-real generalization without accessing target domain data

    Xiangyu Yue, Yang Zhang, Sicheng Zhao, Alberto Sangiovanni-Vincentelli, Kurt Keutzer, and Boqing Gong. Domain randomization and pyramid consistency: Simulation-to-real generalization without accessing target domain data. In ICCV, 2019. 3, 13

  82. [90]

    Style-hallucinated dual consistency learning for domain generalized semantic segmentation

    Yuyang Zhao, Zhun Zhong, Na Zhao, Nicu Sebe, and Gim Hee Lee. Style-hallucinated dual consistency learning for domain generalized semantic segmentation. In ECCV,

  83. [91]

    Adversarial style augmentation for domain general- ized urban-scene segmentation

    Zhun Zhong, Yuyang Zhao, Gim Hee Lee, and Nicu Sebe. Adversarial style augmentation for domain general- ized urban-scene segmentation. In NeurIPS, 2022. 3, 5, 13

  84. [92]

    ibot: Image bert pre-training with online tokenizer

    Jinghao Zhou, Chen Wei, Huiyu Wang, Wei Shen, Cihang Xie, Alan Yuille, and Tao Kong. ibot: Image bert pre-training with online tokenizer. arXiv:2111.07832, 2021. 3

  85. [93]

    Detrs with col- laborative hybrid assignments training

    Zhuofan Zong, Guanglu Song, and Yu Liu. Detrs with col- laborative hybrid assignments training. In ICCV, 2023. 6, 7, 9 19

Pith tools

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