Pith. sign in

REVIEW 2 major objections 7 minor 50 references

Prototype-Anchored Generalized Manifold Regression for Unknown-Domain Object Detection

T0 review · 2 major / 7 minor · reviewed 2026-07-09 · glm-5.2

Pith's one-line read Stop simulating every weather—learn to pull features home instead

desk verdict Solid empirical method with overstated theoretical framing; the 'manifold regression' is prototype-based metric learning, and the rectifier is trained on specific perturbations rather than a general geometric correction rule. read the letter →

arxiv 2607.07192 v1 pith:QLV7OB5Z submitted 2026-07-08 cs.CV

classification cs.CV
keywords single-domaingeneralizationobjectdetectionmanifoldregressionprototypeanchoringdomainshiftfeaturerectificationchain-of-thoughtdiffusionmodels
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 dominant strategy for training object detectors that generalize to unseen domains—exhaustively simulating every possible visual variation via data augmentation or text prompts—is fundamentally limited because finite simulation can never cover infinite real-world shifts. Instead, the authors propose that semantic features for a given object class naturally cluster on a compact, low-dimensional manifold in feature space, and that domain shifts (bad weather, artistic style changes, low light) push samples off this manifold. The key insight is that robust generalization should be reformulated as learning a rectification operator: a parametric mapping that pulls deviant, off-manifold features back toward class-specific prototype neighborhoods on the source semantic manifold. The paper instantiates this through a two-stage closed loop: a Visual-Text Dual Chain-of-Thought module generates structured off-manifold hard examples by coupling text-guided global style evolution with diffusion-based local structural perturbation, and a Class-Specific Prototype Anchoring mechanism trains a regression operator to project these deviant features back toward stable semantic anchors. The central claim is that this simulate-to-deviate, regress-to-rectify paradigm yields superior generalization compared to pure simulation expansion, because it learns a geometric correction rule rather than enumerating perturbation types.

What carries the argument

The central mechanism is a parametric rectification operator H that maps ROI-level instance features into a normalized prototype space where class-specific prototypes serve as manifold anchors. Off-manifold samples are generated by a Dual Chain-of-Thought module: a Textual Chain uses hierarchical CLIP-encoded text prompts (word→phrase→sentence) to predict AdaIN modulation parameters that evolve global style statistics, while a Visual Chain applies Gaussian-blur forward diffusion on content features followed by text-guided reverse diffusion to produce restored-yet-shifted local structure. The regression objective combines a global contrastive loss pulling rectified features toward their class

What would settle it

The central claim would be falsified if detectors trained with the manifold regression approach showed no improvement over pure simulation-driven methods on genuinely unseen domain shifts, or if the learned rectification operator failed to pull features back toward prototypes for perturbation types not used during training (e.g., if training used blur and text-based style shifts but the operator failed on noise-based or occlusion-based corruptions). The t-SNE visualizations in the paper provide qualitative support, but a quantitative test of rectification success rate across held-out pert

Watch

Extended reading notes

Core claim

The paper discovers that reformulating single-domain generalized object detection as a manifold regression problem—where the model learns to rectify off-manifold features back toward class-specific prototype neighborhoods rather than trying to simulate every possible domain shift—produces measurable generalization gains across adverse weather, real-to-art transfer, and zero-shot semantic segmentation benchmarks. The mechanism carrying the argument is the closed loop between structured off-manifold sample generation (combining hierarchical text-guided style evolution with diffusion-based structural perturbation) and prototype-anchored feature regression, which together teach the detector acor

Load-bearing premise

The framework assumes that source-domain instance features form a compact, low-dimensional semantic manifold and that the specific combination of text-guided style evolution and Gaussian-blur diffusion generates off-manifold samples that adequately resemble real-world domain shifts. If the generated perturbations do not structurally match actual domain variations, or if source features do not form a compact manifold, the learned rectification operator may not generalize to真正的

Editorial extensions

If this is right

  • If the manifold regression paradigm is correct, then future domain generalization methods need not attempt exhaustive simulation of target domains but can instead focus on learning stable correction operators that work regardless of the specific perturbation type encountered.
  • The simulate-to-rectify closed loop could be applied to other vision tasks beyond object detection—such as image segmentation, depth estimation, or video understanding—where domain shift degrades feature representations.
  • The class-specific prototype anchoring mechanism suggests that maintaining explicit semantic anchors in feature space may be more robust than implicit alignment methods, particularly under compound domain shifts involving both global style and local structural degradation.
  • The separation of textual (global semantic) and visual (local structural) perturbation chains implies that these two axes of domain shift are partially independent and require complementary simulation strategies.

Reading between the lines

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

  • If the learned rectification operator truly generalizes to unseen shifts, it should also handle adversarial perturbations and corruption types not present in training—a testable extension the paper does not explore but that would strengthen the manifold regression claim.
  • The reliance on Gaussian blur as the visual corruption operator may limit the diversity of structural degradation modes the model learns to correct; replacing blur with other corruption operators (noise, occlusion, compression artifacts) could test whether the rectification operator generalizes across corruption families or is specialized to blur-like degradation.
  • The prototype anchoring mechanism implicitly assumes that intra-class variation in the source domain is representative of legitimate semantic variation; if source data is biased or long-tailed, the learned prototypes may serve as poor anchors for rare subclasses, potentially degrading rectification for those categories.
  • The framework's claim that semantic features lie on a compact manifold could be empirically tested by measuring the intrinsic dimensionality of source-domain ROI features and comparing it to the effective dimensionality under domain shift—if the manifold hypothesis holds, shifted features should occupy a higher-dimensional space that the rectification operator compresses back down.
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

2 major / 7 minor

Summary. This paper proposes MR-DCoT for Single-Domain Generalized Object Detection (Single-DGOD). The core idea is to reformulate generalization as a manifold regression problem: rather than exhaustively simulating domain shifts, the method learns a rectification operator that projects off-manifold features back toward class-specific prototype neighborhoods on the source semantic manifold. The framework has three components: (1) Disentangled Feature Embedding, which separates style and content; (2) a Visual-Text Dual Chain-of-Thought (Dual-CoT) module that generates structured off-manifold hard examples by combining VLM-guided textual style evolution with diffusion-based (Gaussian blur) structural perturbation; and (3) Class-Specific Prototype Anchoring, which provides manifold anchors for a hierarchical regression objective (global prototype contrast + local paired-feature regression). The method is evaluated on adverse-weather detection, real-to-art generalization, and zero-shot semantic segmentation, across multiple detection architectures (Faster R-CNN, YOLO, DiffusionDet, GLIP, DINO). Ablation studies isolate the contributions of each component.

Significance. The paper makes a conceptually interesting reframing of the Single-DGOD problem: instead of treating domain generalization as broader simulation coverage, it explicitly trains a feature-rectification operator. The empirical evaluation is extensive, spanning five detection frameworks and three task benchmarks, with consistent improvements over prior methods including the authors' own SE-COT. The ablation studies (Tables XI-XVIII) are thorough, examining component interactions, prototype configurations, diffusion steps, injection layers, and hyperparameter sensitivity. The prompt-robustness experiment (Table XVI) addresses a reasonable concern about LLM dependency. The framework is architecturally agnostic, which is a practical strength. However, the central conceptual claim—that the learned operator constitutes a general geometric correction rule rather than an inversion of the specific Dual-CoT perturbation pipeline—is not formally established and is only partially supported by the experiments.

major comments (2)
  1. Sec. I and Sec. III-D frame the contribution as a paradigm shift from 'simulation' to 'regression,' claiming the learned rectification operator H(·) masters a general geometric correction rule rather than inverting a specific perturbation generator. However, H(·) is trained exclusively on Dual-CoT-generated off-manifold samples (AdaIN style transfer + Gaussian blur diffusion) with paired clean-source supervision via L_local (Eq. 24). There is no formal argument or experimental evidence that the learned inverse mapping generalizes beyond the Dual-CoT perturbation distribution. A direct test would be: train the rectifier using only one perturbation type (e.g., fog-style textual prompts + blur) and evaluate on other unseen shifts (rain, night). Table XVI varies textual descriptions but not structural perturbation types. Without such a cross-shift generalization test, the claim that the rect
  2. Table XI ablation reveals that DFE + Dual-CoT without MR already yields the bulk of the improvement (baseline 43.4 → 53.1 on Night Sunny; 27.5 → 34.2 on Night Rainy), with MR adding 1.5–3.6 mAP on top. This suggests the simulation/augmentation strategy, not the rectification paradigm, drives most of the gain. The paper should more carefully contextualize this: if the rectification operator's marginal contribution is modest, the framing as a fundamental paradigm shift (Sec. I, Sec. III-D) may be overstated relative to what the ablation supports. The authors should either (a) provide experiments showing MR's benefit grows under more severe or diverse shifts where augmentation alone fails, or (b) moderate the paradigm-shift language to match the empirical marginal contribution of MR.
minor comments (7)
  1. Sec. III-C, Eqs. (16)–(18): The semantic manifold is defined as the union of prototype-centered neighborhoods, with radius r_c set to the 95th percentile of source-feature distances. This is a practical heuristic, but the paper uses manifold terminology (geodesic, Riemannian) in Sec. II-C without connecting it to the actual implementation. Clarifying that the 'manifold' here is an operational approximation (prototype neighborhoods in feature space) rather than a formal Riemannian structure would improve precision.
  2. Sec. III-B2, Eq. (12): The dynamic blur schedule adjusts s_v^t based on local mean and variance via an MLP g(·). It is unclear whether g(·) is trained jointly with the detector or frozen. If trained jointly, the forward diffusion process changes during training, which could affect the stability of the off-manifold samples. Please clarify.
  3. Table XIV: The full model adds 1.8M parameters and ~10ms inference time over the baseline. The paper states Dual-CoT operates only during training, but MR adds inference overhead via prototype-guided feature enhancement (Eq. 22). This overhead should be discussed in the context of real-time detection applications, particularly for the YOLO framework.
  4. Sec. III-B1: The paper states GPT is used offline to generate keyword groups, which are then fixed. However, the exact prompts used to query GPT and the full list of generated keyword groups are not provided in the manuscript or supplementary. For reproducibility, these should be included (or released with code), since the textual chain depends on the generated vocabulary.
  5. Fig. 2: The notation in the overview figure uses symbols (e.g., question marks for prototype labels) that are not immediately clear. The figure would benefit from clearer labeling matching the notation used in the equations.
  6. Sec. III-C, Eq. (23): The contrastive formulation for L_global pushes H(z_off) toward P_gt and away from other prototypes. However, since prototypes are learned from source features, this objective may bias the rectified features toward source-domain statistics. A brief discussion of whether this introduces any source-domain overfitting risk, and how the local regression term (Eq. 24) counterbalances it, would help.
  7. The paper references [14] (Li and He, 2025) for the manifold hypothesis and denoising perspective. The connection between this work's discriminative rectification and the generative denoising framework in [14] is mentioned but not deeply explored. A more precise articulation of what is borrowed beyond shared motivation would strengthen the theoretical positioning.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity found; the derivation chain is self-contained with one minor self-citation to a preliminary conference version.

full rationale

The paper's derivation chain is self-contained and does not reduce to its inputs by construction. The three main components operate on distinct inputs: (1) Disentangled Feature Embedding decouples source features into style/content via lightweight encoders with orthogonality and consistency losses (Eqs. 1-4); (2) Dual-CoT generates off-manifold samples by evolving style via text-conditioned AdaIN (Eq. 8) and perturbing content via Gaussian-blur diffusion with text-guided reverse restoration (Eqs. 10-15); (3) Manifold Regression learns prototypes from source features via soft assignment and residual aggregation (Eqs. 19-21), then trains a rectification operator H(·) to pull off-manifold features toward class-specific prototype neighborhoods and paired clean source features (Eqs. 23-25). No equation reduces to another by definition. The regression targets (prototypes, clean source features) are derived from source-domain data, while the regression inputs (off-manifold features) are generated by a separate process (Dual-CoT). The skeptic's concern that H(·) merely inverts the Dual-CoT perturbation distribution is a generalization/correctness risk, not a circularity issue—the paper does not claim H(·) is derived from or defined by the perturbation generator; it is a learned operator evaluated empirically on genuinely unseen target domains. The self-citation to SE-COT [9] (same authors) is acknowledged as a preliminary conference version and is not load-bearing for the central manifold regression claim, which is new to this paper. The manifold hypothesis is cited from Li et al. [14] (different authors), providing external theoretical grounding. Score 1 reflects the minor, non-load-bearing self-citation to SE-COT for the textual chain component.

Assumptions & free parameters 7 free parameters · 3 assumptions · 2 invented entities

The paper relies on standard domain assumptions (manifold hypothesis, disentanglement) and introduces several empirically tuned hyperparameters. The 'semantic manifold' is a conceptual construct approximated by learned prototypes, which are standard in metric learning.

free parameters (7)
  • lambda_sc = 0.5
    Weight for style consistency loss, chosen empirically.
  • alpha = 0.5
    Dynamic blur magnitude control factor, chosen empirically.
  • tau = 0.1
    Temperature for contrastive loss, chosen empirically.
  • lambda_loc = 0.5
    Weight for local regression loss, chosen empirically.
  • q = 0.95
    Percentile for semantic tolerance radius, default value.
  • T = 8
    Diffusion steps, chosen empirically via ablation.
  • m_r = 5
    Sampled keyword number, chosen empirically.
assumptions (3)
  • domain assumption Manifold Hypothesis
    The paper assumes semantic features lie on a compact, low-dimensional manifold (Section I, III-C). This is a standard assumption in deep learning but is not proven for the specific feature spaces used here.
  • domain assumption Style-Content Disentanglement
    The method assumes backbone features can be decoupled into orthogonal style and content components (Section III-A).
  • ad hoc to paper Off-manifold simulation validity
    The paper assumes that the specific combination of AdaIN style transfer and Gaussian blur diffusion generates off-manifold samples that are valid proxies for real-world domain shifts (Section III-B).
invented entities (2)
  • Semantic Manifold
    purpose: Conceptual target for feature regression
    The manifold is not a new physical entity but a geometric metaphor for the class-conditional feature support. It is approximated by prototype neighborhoods.
  • Class-Specific Prototypes independent evidence
    purpose: Anchors for manifold regression
    Learned cluster centers in feature space. Their effectiveness is validated by ablation studies (Table XV) and downstream detection performance.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Prototype-Anchored Generalized Manifold Regression for Unknown-Domain Object Detection." pith.science (2026). https://pith.science/paper/QLV7OB5Z

@misc{pith2026260707192,
  author       = {Pith},
  title        = {Pith review of: Prototype-Anchored Generalized Manifold Regression for Unknown-Domain Object Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QLV7OB5Z}},
  note         = {Machine review of arXiv:2607.07192}
}
read the original abstract

In this paper, we study Single-Domain Generalized Object Detection (Single-DGOD), which aims to transfer a detector trained on a single source domain to multiple unseen domains. Existing methods mainly rely on simulation-driven strategies, such as data augmentation or textual prompts, to enlarge the training distribution. However, finite simulations can hardly cover the dynamic variations of real-world scenarios, often causing overfitting to synthetic styles and limited robustness to complex structural degradations. Inspired by the manifold hypothesis, we argue that semantic features, despite diverse visual changes, should lie on a compact and stable low-dimensional manifold. Therefore, robust generalization requires rectifying deviant samples back to this semantic manifold, rather than exhaustively simulating external perturbations. To this end, we propose Manifold Regression with Visual-Text Dual Chain-of-Thought (MR-DCoT), which formulates unknown-domain generalization as a manifold regression problem. MR-DCoT first uses a Visual-Text Dual Chain-of-Thought module to combine VLM-guided semantic evolution with diffusion-based structural perturbation, generating structured off-manifold hard examples. It then introduces Class-Specific Prototype Anchoring to learn a rectification operator that projects deviant features toward the source semantic manifold. By integrating outlier generation and semantic correction into a closed loop, MR-DCoT effectively narrows the distribution gap and improves robustness under unseen shifts. Extensive experiments on three complementary benchmarks, including adverse-weather detection, real-to-art generalization, and zero-shot semantic segmentation, demonstrate the effectiveness and versatility of our method.

Figures

Figures reproduced from arXiv: 2607.07192 by the authors.

Figure 1
Figure 1. Manifold Regression with Visual-Text Dual Chain-of-Thought (MR-DCoT) for detecting unknown-domain objects. The core of MR-DCoT lies in the duality of simulation and rectification: (a) Off-Manifold Generation employs a Visual-Text Dual Chain-of-Thought to simulate structured off-manifold outliers, driving features away from the source distribution; (b) Manifold Regression learns a robust rectification mechanism that … view at source ↗
Figure 2
Figure 2. Overview of the Manifold Regression with Visual-Text Dual Chain-of-Thought (MR-DCoT) framework. First, the input image features are decoupled into style (Fs) and content (Fc) components via the Disentangled Feature Embedding module, ensuring semantic integrity through contrastive and consistency losses. Next, the Visual-Text Dual-CoT Module is employed to generate structured off-manifold outliers. The Textual Chain … view at source ↗
Figure 3
Figure 3. Complex Style Evolution via Textual Chains. By employing text prompts that progress from simplicity to complexity, the style is continuously evolved and expanded, thereby simulating a wider variety of style features with distinct data distributions. Parameters µt and σt are trained using consistency loss to align textual features with visual features. pervision for learning a robust regression mechanism, we pro￾pose… view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: Qualitative Results: Detection results under different weather conditions. The four rows display the results from C-Gap [1], PDOC [10], SE-COT [9], and our method, respectively. Objects highlighted in red boxes indicate missed detections or misclassifications. TABLE II…
Figure 5
Figure 5. Figure 5: Hyperparameter sensitivity analysis. (a) Textual Chain Hierarchy: Performance peaks at the sentence level (Level 3), indicating an optimal balance of semantic guidance. (b) Visual Chain Diffusion Steps T: Detection accuracy increases with T until saturation at T = 8, d…
Figure 6
Figure 6. Figure 6: Qualitative detection results across five distinct weather scenarios. We showcase the inference performance on the source domain (Day Clear) and four unseen target domains (Night Sunny, Night Rainy, Dusk Rainy, and Day Foggy). Despite severe illumination changes, rain …
Figure 7
Figure 7. Figure 7: More Hyperparameter Sensitivity Analysis. The impact of dynamic blur magnitude α, regression weight λloc, and temperature hyperparameter τ on detection performance (mAP) across Night Sunny and Day Foggy domains. hindering the Manifold Regression module from recovering …
Figure 8
Figure 8. Figure 8: Qualitative analysis of Manifold Regression. We visualize the activation and feature maps to evaluate the rectification capability. Compared to the baseline without Manifold Regression (b, d), our method (c, e) significantly suppresses background noise (e.g., rain stre…
Figure 9
Figure 9. Figure 9: Qualitative analysis of Textual Chain Hierarchy. We visualize the feature maps guided by different levels of the textual chain. Compared to the One-Step prompt (b), the Three-Level (Sentence-level) chain (d) significantly suppresses background noise and focuses activat…
Figure 10
Figure 10. Figure 10: t-SNE Visualization of Off-Manifold Simulation and Manifold Regression. The two subplots show that Dual-CoT generates off-manifold features under different textual guidance, while Manifold Regression rectifies these deviant features back toward the source semantic reg…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

50 extracted references · 50 canonical work pages

  1. [1]

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

    V . Vidit, M. Engilberge, and M. Salzmann, “Clip the gap: A single domain generalization approach for object detection,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 3219–3229

  2. [2]

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

    M. S. Danish, M. H. Khan, M. A. Munir, M. S. Sarfraz, and M. Ali, “Improving single domain-generalized object detection: A focus on diversification and alignment,” inProceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, 2024, pp. 17 732– 17 742

  3. [3]

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

    Y . Liu, S. Zhou, X. Liu, C. Hao, B. Fan, and J. Tian, “Unbiased faster r-cnn for single-source domain generalized object detection,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 28 838–28 847

  4. [4]

    Behind every domain there is a shift: Adapting distortion-aware vision transformers for panoramic semantic segmentation,

    J. Zhang, K. Yang, H. Shi, S. Reiß, K. Peng, C. Ma, H. Fu, P. H. Torr, K. Wang, and R. Stiefelhagen, “Behind every domain there is a shift: Adapting distortion-aware vision transformers for panoramic semantic segmentation,”IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024

  5. [5]

    Tib: Detecting unknown objects via two-stream information bottleneck,

    A. Wu and C. Deng, “Tib: Detecting unknown objects via two-stream information bottleneck,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 46, no. 1, pp. 611–625, 2023

  6. [6]

    Vector-decomposed disentanglement for domain-invariant object detection,

    A. Wu, R. Liu, Y . Han, L. Zhu, and Y . Yang, “Vector-decomposed disentanglement for domain-invariant object detection,” inProceedings of the IEEE/CVF International Conference on Computer Vision, 2021, pp. 9342–9351

  7. [7]

    Universal-prototype enhancing for few-shot object detection,

    A. Wu, Y . Han, L. Zhu, and Y . Yang, “Universal-prototype enhancing for few-shot object detection,” inProceedings of the IEEE/CVF Inter- national Conference on Computer Vision, 2021, pp. 9567–9576

  8. [8]

    Poda: Prompt-driven zero-shot domain adaptation,

    M. Fahes, T.-H. Vu, A. Bursuc, P. P ´erez, and R. de Charette, “Poda: Prompt-driven zero-shot domain adaptation,” inProceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 18 623–18 633

Show all 50 references
  1. [9]

    Style evolving along chain-of-thought for unknown-domain object detection,

    Z. Zhang, A. Wu, and Y . Han, “Style evolving along chain-of-thought for unknown-domain object detection,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2025, pp. 14 225–14 234

  2. [10]

    Prompt-driven dynamic object- centric learning for single domain generalization,

    D. Li, A. Wu, Y . Wang, and Y . Han, “Prompt-driven dynamic object- centric learning for single domain generalization,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 17 606–17 615

  3. [11]

    Percept, memory, and imagine: World feature sim- ulating for open-domain unknown object detection,

    A. Wu and C. Deng, “Percept, memory, and imagine: World feature sim- ulating for open-domain unknown object detection,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2025, pp. 4682–4691

  4. [12]

    Source-free domain adaptation with frozen multimodal foundation model,

    S. Tang, W. Su, M. Ye, and X. Zhu, “Source-free domain adaptation with frozen multimodal foundation model,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 23 711–23 720

  5. [13]

    Towards ood object detection with unknown- concept guided feature diffusion,

    A. Wu and C. Deng, “Towards ood object detection with unknown- concept guided feature diffusion,”IEEE Transactions on Pattern Anal- ysis and Machine Intelligence, vol. 47, no. 11, pp. 9798–9812, 2025

  6. [14]

    Back to basics: Let denoising generative models denoise,

    T. Li and K. He, “Back to basics: Let denoising generative models denoise,”arXiv preprint arXiv:2511.13720, 2025

  7. [15]

    Deep feature deblurring diffusion for detecting out-of-distribution objects,

    A. Wu, D. Chen, and C. Deng, “Deep feature deblurring diffusion for detecting out-of-distribution objects,” inProceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 13 381–13 391

  8. [16]

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

    A. Wu and C. Deng, “Single-domain generalized object detection in urban scene via cyclic-disentangled self-distillation,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 847–856

  9. [17]

    Robust domain adaptive object detection with unified multi-granularity alignment,

    L. Zhang, W. Zhou, H. Fan, T. Luo, and H. Ling, “Robust domain adaptive object detection with unified multi-granularity alignment,” IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024

  10. [18]

    Single-domain generalized object detection with frequency whitening and contrastive learning,

    X. Guo, C. Liu, X. Qian, Z. Wang, X. Feng, and Y . Xue, “Single-domain generalized object detection with frequency whitening and contrastive learning,”IEEE Transactions on Multimedia, pp. 1–14, 2025

  11. [19]

    A comprehensive survey on source-free domain adaptation,

    J. Li, Z. Yu, Z. Du, L. Zhu, and H. T. Shen, “A comprehensive survey on source-free domain adaptation,”IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024

  12. [20]

    G-nas: Generalizable neural architecture search for single domain generalization object detection,

    F. Wu, J. Gao, L. Hong, X. Wang, C. Zhou, and N. Ye, “G-nas: Generalizable neural architecture search for single domain generalization object detection,” inProceedings of the AAAI Conference on Artificial Intelligence, vol. 38, no. 6, 2024, pp. 5958–5966

  13. [21]

    Learning transferable visual models from natural language supervision,

    A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sastry, A. Askell, P. Mishkin, J. Clarket al., “Learning transferable visual models from natural language supervision,” inInternational Conference on Machine Learning. PMLR, 2021, pp. 8748–8763

  14. [22]

    Deep defocus map estimation using domain adaptation,

    J. Lee, S. Lee, S. Cho, and S. Lee, “Deep defocus map estimation using domain adaptation,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019, pp. 12 222–12 230. JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 18

  15. [23]

    Geodesic regression and the theory of least squares on riemannian manifolds,

    P. Thomas Fletcher, “Geodesic regression and the theory of least squares on riemannian manifolds,”International Journal of Computer Vision, vol. 105, no. 2, pp. 171–185, 2013

  16. [24]

    Contractive auto-encoders: Explicit invariance during feature extraction,

    S. Rifai, P. Vincent, X. Muller, X. Glorot, and Y . Bengio, “Contractive auto-encoders: Explicit invariance during feature extraction,” inProceed- ings of the 28th international conference on international conference on machine learning, 2011, pp. 833–840

  17. [25]

    A connection between score matching and denoising au- toencoders,

    P. Vincent, “A connection between score matching and denoising au- toencoders,”Neural computation, vol. 23, no. 7, pp. 1661–1674, 2011

  18. [26]

    Show, attend and tell: Neural image caption generation with visual attention,

    K. Xu, J. Ba, R. Kiros, K. Cho, A. Courville, R. Salakhudinov, R. Zemel, and Y . Bengio, “Show, attend and tell: Neural image caption generation with visual attention,” inInternational Conference on Machine Learning. PMLR, 2015, pp. 2048–2057

  19. [27]

    Gpt-4 technical report,

    J. Achiam, S. Adler, S. Agarwal, L. Ahmad, I. Akkaya, F. L. Aleman, D. Almeida, J. Altenschmidt, S. Altman, S. Anadkatet al., “Gpt-4 technical report,”arXiv preprint arXiv:2303.08774, 2023

  20. [28]

    Adain-based tunable cyclegan for efficient unsu- pervised low-dose ct denoising,

    J. Gu and J. C. Ye, “Adain-based tunable cyclegan for efficient unsu- pervised low-dose ct denoising,”IEEE Transactions on Computational Imaging, vol. 7, pp. 73–85, 2021

  21. [29]

    Motiondiffuse: Text-driven human motion generation with diffusion model,

    M. Zhang, Z. Cai, L. Pan, F. Hong, X. Guo, L. Yang, and Z. Liu, “Motiondiffuse: Text-driven human motion generation with diffusion model,”IEEE Transactions on Pattern Analysis and Machine Intelli- gence, vol. 46, no. 6, pp. 4115–4128, 2024

  22. [30]

    Diffusion models in vision: A survey,

    F.-A. Croitoru, V . Hondru, R. T. Ionescu, and M. Shah, “Diffusion models in vision: A survey,”IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 45, no. 9, pp. 10 850–10 869, 2023

  23. [31]

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

    N. Inoue, R. Furuta, T. Yamasaki, and K. Aizawa, “Cross-domain weakly-supervised object detection through progressive domain adap- tation,” inProceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018, pp. 5001–5009

  24. [32]

    The pascal visual object classes (voc) challenge,

    M. Everingham, L. Van Gool, C. K. Williams, J. Winn, and A. Zisser- man, “The pascal visual object classes (voc) challenge,”International Journal of Computer Vision, vol. 88, pp. 303–338, 2010

  25. [33]

    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,” inProceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016, pp. 3213–3223

  26. [34]

    Acdc: The adverse conditions dataset with correspondences for semantic driving scene understanding,

    C. Sakaridis, D. Dai, and L. Van Gool, “Acdc: The adverse conditions dataset with correspondences for semantic driving scene understanding,” inProceedings of the IEEE/CVF International Conference on Computer Vision, 2021, pp. 10 765–10 775

  27. [35]

    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,” inEuropean Conference on Computer Vision. Springer, 2016, pp. 102–118

  28. [36]

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

    S. Ren, K. He, R. Girshick, and J. Sun, “Faster r-cnn: Towards real-time object detection with region proposal networks,”Advances in Neural Information Processing Systems, vol. 28, 2015

  29. [37]

    Switchable whitening for deep representation learning,

    X. Pan, X. Zhan, J. Shi, X. Tang, and P. Luo, “Switchable whitening for deep representation learning,” inProceedings of the IEEE/CVF International Conference on Computer Vision, 2019, pp. 1863–1871

  30. [38]

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

    S. Choi, S. Jung, H. Yun, J. T. Kim, S. Kim, and J. Choo, “Robustnet: Improving domain generalization in urban-scene segmentation via in- stance selective whitening,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021, pp. 11 580–11 590

  31. [39]

    Srcd: Se- mantic reasoning with compound domains for single-domain generalized object detection,

    Z. Rao, J. Guo, L. Tang, Y . Huang, X. Ding, and S. Guo, “Srcd: Se- mantic reasoning with compound domains for single-domain generalized object detection,”IEEE Transactions on Neural Networks and Learning Systems, vol. 36, no. 7, pp. 12 497–12 506, 2024

  32. [40]

    Yolov10: Real-time end-to-end object detection,

    A. Wang, H. Chen, L. Liu, K. Chen, Z. Lin, J. Hanet al., “Yolov10: Real-time end-to-end object detection,”Advances in Neural Information Processing Systems, vol. 37, pp. 107 984–108 011, 2024

  33. [41]

    Diffusiondet: Diffusion model for object detection,

    S. Chen, P. Sun, Y . Song, and P. Luo, “Diffusiondet: Diffusion model for object detection,” inProceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 19 830–19 843

  34. [42]

    Grounded language-image pre- training,

    L. H. Li, P. Zhang, H. Zhang, J. Yang, C. Li, Y . Zhong, L. Wang, L. Yuan, L. Zhang, J.-N. Hwanget al., “Grounded language-image pre- training,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 10 965–10 975

  35. [43]

    Phrase grounding-based style transfer for single- domain generalized object detection,

    H. Li, W. Wang, C. Wang, M. Wang, X. Zhang, L. Lan, X. Liu, K. Li, and X. Cao, “Phrase grounding-based style transfer for single- domain generalized object detection,”IEEE Transactions on Circuits and Systems for Video Technology, 2025

  36. [44]

    Dino: Detr with improved denoising anchor boxes for end-to-end object detection,

    H. Zhang, F. Li, S. Liu, L. Zhang, H. Su, J. Zhu, L. Ni, and H.-Y . Shum, “Dino: Detr with improved denoising anchor boxes for end-to-end object detection,” inThe Eleventh International Conference on Learning Representations, 2023

  37. [45]

    Towards single- source domain generalized object detection via causal visual prompts,

    C. Li, H. Xu, C. Gao, Z. Wang, Y . Liu, and X. Zhu, “Towards single- source domain generalized object detection via causal visual prompts,” in The Thirty-ninth Annual Conference on Neural Information Processing Systems, 2025

  38. [46]

    Towards robust object detection invariant to real-world domain shifts,

    Q. Fan, M. Segu, Y .-W. Tai, F. Yu, C.-K. Tang, B. Schiele, and D. Dai, “Towards robust object detection invariant to real-world domain shifts,” inThe Eleventh International Conference on Learning Representations. OpenReview, 2023

  39. [47]

    Clipstyler: Image style transfer with a single text condition,

    G. Kwon and J. C. Ye, “Clipstyler: Image style transfer with a single text condition,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 18 062–18 071

  40. [48]

    Unified language-driven zero- shot domain adaptation,

    S. Yang, Z. Tian, L. Jiang, and J. Jia, “Unified language-driven zero- shot domain adaptation,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 23 407–23 415

  41. [49]

    Openai gpt-5 system card,

    A. Singh, A. Fry, A. Perelman, A. Tart, A. Ganesh, A. El-Kishky, A. McLaughlin, A. Low, A. Ostrow, A. Ananthramet al., “Openai gpt-5 system card,”arXiv preprint arXiv:2601.03267, 2025

  42. [50]

    Gemini: a family of highly capable multimodal models,

    G. Team, R. Anil, S. Borgeaud, J.-B. Alayrac, J. Yu, R. Soricut, J. Schalkwyk, A. M. Dai, A. Hauth, K. Millicanet al., “Gemini: a family of highly capable multimodal models,”arXiv preprint arXiv:2312.11805, 2023. Zihao Zhangreceived his M.Eng. degree from the College of Intell...

Pith tools

Reviewed July 9, 2026 · model on record in the stance chip above.