Pith. sign in

REVIEW 2 major objections 5 minor 81 references

CamoDreamer claims that camouflaged-image generation is best done by decoupling object and background control streams, then blending them per frequency, and reports a 15.54-point FID improvement over the prior state of the art.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-01 17:03 UTC pith:JR332XMM

load-bearing objection A well-ablated, plausible architecture whose headline FID gain can't be trusted until the evaluation set-disjointness is clarified. the 2 major comments →

arxiv 2607.17768 v1 pith:JR332XMM submitted 2026-07-20 cs.CV

To Blend In, First Decouple: Rethinking Camouflage Image Generation via Context-Decoupled Representations

Wenzhuang Wang , Yifan Zhao , Mingcan Ma , Yunlong Che , Haoran Chen , Ming Liu , Jia Li This is my paper
classification cs.CV
keywords camouflage image generationcontext-decoupled generationdiffusion modelsconditional image synthesisforeground-background disentanglementfrequency-domain blendingsynthetic data augmentationcamouflage object detection
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper argues that existing camouflage image generation fails because object and background conditions are entangled in one guidance space, causing 'cross-context representation leakage' — background artifacts and color mismatches. It proposes CamoDreamer, a 'decouple-then-blend' diffusion pipeline that splits latent camouflage features into separate object and background streams, each with its own attention and mask, and recombines them with a frequency-adaptive blend that keeps high-frequency object texture and low-frequency background structure. On the LAKE-RED benchmark the authors report that CamoDreamer beats all 11 prior methods, lowering overall FID from 52.87 to 37.33 and KID from 0.0169 to 0.0159. They also show that a camouflage detector trained on CamoDreamer's synthetic images performs better on real benchmarks than detectors trained on other synthetic sets, evidence that the generated data can relieve camouflage data scarcity. The core truth, if confirmed, is that controlling object and background separately is the right abstraction for camouflaged-image synthesis.

Core claim

The central claim is that the bottleneck in camouflage image generation is not generator capacity but entangled conditioning: background cues leak into foreground rendering and vice versa, producing artifacts and mismatches. CamoDreamer separates latent features into object and background streams, each with its own masked cross-attention guided by contrast-aware retrieved-background cues, then recombines them with a learnable high/low-frequency blend. On the LAKE-RED benchmark it reports the lowest FID (37.33) and KID (0.0159), a 15.54-point FID gain over the previous best, and detectors trained on its synthetic output gain on real camouflage benchmarks.

What carries the argument

The engine is a 'decouple, then blend' architecture: a Contrast-aware Contextual Bridge derives dual guidance from the difference between foreground and retrieved-background features; Context-Decoupled Assimilation Streams run separate masked cross-attention for object and background latents so each attends only to its own conditioning; a Frequency-Adaptive Contextual Blend fuses them with learned high/low-pass Fourier masks and a weighted sum. A target-prior re-weighting loss amplifies loss on small objects. Each module exists to stop one context's features from leaking into the other's rendering.

Load-bearing premise

The quantitative case rests on the assumption that the 5,066 real COD10K images used to compute FID/KID are disjoint from the 4,040 training images that form the retrieval knowledge base and from the 6,473 evaluation foregrounds used as conditioning inputs; the paper never states this, and overlap would bias scores toward re-rendering the input rather than measuring camouflage quality.

What would settle it

Look for overlap between the 5,066 COD10K reference images and the 4,040-image LAKE-RED training set; if overlap exists, recompute FID/KID after excluding those references. Additionally, run an ablation that masks out the input foreground region in the generated images before computing FID; if the 15.54-point advantage over the next best method shrinks sharply, the gain may come from outputting a near-copy of the target rather than from context decoupling.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • If the reported numbers hold, context-decoupled conditioning becomes the top-scoring approach on the LAKE-RED benchmark, with FID falling from 52.87 to 37.33 and KID from 0.0169 to 0.0159.
  • Training a camouflage detector on CamoDreamer-synthesized images yields consistent gains over other synthetic training sets on CAMO, COD10K, and NC4K (e.g., +0.053 S-measure on CAMO relative to LAKE-RED synthetic data), suggesting synthetic data can ease the camouflage annotation bottleneck.
  • Because only the U-Net middle and decoder layers are fine-tuned (172.22M trainable parameters), the approach is comparatively cheap to deploy on a single 24GB GPU.
  • The decouple-then-blend design is claimed to transfer across camouflaged, salient, and general object categories, narrowing the FID gap between them.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • A decisive test of the decoupling hypothesis is to hold inputs fixed and compare the two-stream module stack against a single-stream model that receives the same retrieved backgrounds and masks on a reference set guaranteed disjoint from the knowledge base; if the FID gap persists, decoupling — not retrieval or input copying — is responsible.
  • The paper's own supplementary limitation admits local distortion on complex, salient objects; an extension that injects shape constraints into the object stream without re-entangling background cues would test whether the remaining error is inherent to decoupling or fixable.
  • The frequency-split blending idea may transfer to other 'hide-in-plain-sight' tasks such as adversarial patch design or privacy-preserving object removal, where structural coherence between foreground and background is the core requirement.
  • The reference-set composition deserves scrutiny: if the 5,066 COD10K real images used for FID overlap the retrieval knowledge base or conditioning inputs, part of the reported gain may reflect appearance copying rather than camouflage fidelity.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

2 major / 5 minor

Summary. The paper proposes CamoDreamer, a camouflage image generation (CIG) method built on Stable Diffusion v1.5. The method introduces a 'decouple-then-blend' paradigm: a Contrast-aware Contextual Bridge retrieves background anchors and builds contrast-aware dual conditional guidance; Context-Decoupled Assimilation Streams separate latent camouflage features into object and background attention streams; and a Frequency-Adaptive Contextual Blend merges the decoupled features before decoding. A target-prior re-weighting term upweights foreground reconstruction. The authors report state-of-the-art FID/KID on the LAKE-RED benchmark (overall FID 37.33, KID 0.0159, a 15.54 FID improvement over CT-CIG), better downstream camouflage-object detection when synthetic images are used as training data, and ablations supporting each component.

Significance. If the reported numbers hold, this is a solid empirical contribution to CIG: the architecture is modular, the ablations are reasonably complete, the downstream COD evaluation is useful, and the paper explicitly compares against a non-generative Copy-Paste baseline. The large reported FID margin, the consistent trend across three object subsets, and the downstream gains would make the 'context decoupling' claim credible. The main caveat is that the headline quantitative claim rests on an evaluation protocol whose independence is not established: the FID/KID reference set, the retrieval knowledge base, and the conditioning foregrounds may overlap, which could inflate the reported gains by rewarding input reconstruction rather than camouflage synthesis.

major comments (2)
  1. [§4.1, Supplementary §A, Eqs. (5), (8), (15)] Supplementary §A states FID/KID are computed against 5,066 real COD10K (CAM) images. Section 4.1 states the knowledge base is constructed from the 4,040 LAKE-RED training images, of which 3,040 come from COD10K. The paper never asserts that these sets are disjoint from each other or from the 6,473 'Camouflage Objects' conditioning inputs. If they overlap, the retrieval in Eq. (5) can return the original background of a training image, and Eqs. (8) and (15) encourage the decoder to reproduce the input foreground and that background, so FID/KID would measure reconstruction of the reference set rather than camouflage fidelity. Table 1's 15.54-point overall FID advantage over CT-CIG is then not a valid superiority claim. The authors must prove disjointness and, if needed, recompute all FID/KID values using a reference set excluded from the KB and from the conditioning foregrounds.
  2. [§4.1, Table 7, Supplementary §B.1] All FID/KID values in Tables 1, 6, 7, and 9 are single-run point estimates with no error bars, seeds, or significance tests. More importantly, hyperparameters λ, K, the CCB feature extractor, and the FACB frequency mask are selected on the same LAKE-RED evaluation split that produces the headline Table 1 result. The sensitivity is not small: λ=2 gives overall FID 45.24, and removing the learnable frequency mask gives 39.87. A held-out validation split or multiple-seed reporting is needed before the 'consistently best' claim can be taken at face value.
minor comments (5)
  1. [§4.1] The provenance of the FID/KID reference set and its relationship to the training/KB set should be stated in the main text, not only in Supplementary §A. Please also state explicitly whether any of the 6,473 evaluation foregrounds appear in the 5,066-image reference set.
  2. [Table 3] The total parameter count for CamoDreamer is printed as '1238.45' with no thousands separator; use '1,238.45' or report the breakdown. The 'relatively lightweight' claim should be based on trainable parameters only, which is the more meaningful comparison here.
  3. [§3.4–3.5] Notation is occasionally imprecise: Eq. (8) uses M̄_fg for the downsampled mask but the definition is easy to miss; in Eq. (11) δ([q_fg; q_bg]) and the frequency masks M↑, M↓ should be defined more explicitly; in Eq. (12) clarify the roles of π↑ and π↓.
  4. [Supplementary §B.3] The limitation statement about possible local distortions for salient or structurally complex objects is useful, but it sits in tension with the abstract's claim of 'consistently' superior performance. Add a sentence in the main text acknowledging this boundary condition.
  5. [General] The manuscript cites a project page URL but no code or checkpoint release. Since the evaluation-protocol concern cannot be checked from the paper alone, a code release or a detailed supplementary protocol description would materially strengthen reproducibility.

Circularity Check

0 steps flagged

No significant circularity: CamoDreamer is an empirical system paper; no reported result is forced by the paper's own equations or by load-bearing self-citation.

full rationale

Walking the claimed derivation chain, this is an empirical system paper rather than a formal derivation. The headline claims are that a retrieval-conditioned, masked two-stream diffusion fine-tune (CamoDreamer) achieves lower FID/KID on LAKE-RED and improves downstream COD. The equations in the paper define the architecture and training losses (Eqs. 5-15), but none of them equates an output metric to a fitted parameter by construction. The ablations attribute FID changes to specific modules, and even if hyperparameters such as λ and K were selected on the same benchmark, that is model-selection bias rather than logical circularity. The self-citations (e.g., refs. [3], [55]) appear only in related-work surveys and are not load-bearing for the central claim. The supplementary 'Generative Limitations' section acknowledges open-domain distortions, which is an honest performance caveat rather than an admission of circularity. The skeptically noted possibility that the FID reference set overlaps the retrieval knowledge base or conditioning foregrounds is a potential evaluation-validity concern, but it is not an equation-level identity: the paper does not define correctness in terms of its own inputs, and any such overlap would affect the baselines as well. Under the rule that circularity must be exhibited as a specific reduction, no such step can be quoted from this manuscript.

Axiom & Free-Parameter Ledger

3 free parameters · 5 axioms · 0 invented entities

No new physical entities are introduced; the three modules (CCB, CDAS, FACB) are architectural components, not independent postulations requiring external falsification. The load-bearing assumptions are the validity of the SDv1.5/VQ-VAE substrate, the retrieval quality, the foreground/background frequency dichotomy, mask fidelity, and the integrity of the FID reference set.

free parameters (3)
  • λ (target prior re-weight factor) = 3
    Eq. 14; selected by ablation on LAKE-RED evaluation FID (Table 7); controls foreground/background loss balance and directly affects reported FID.
  • K (number of background anchors) = 3
    Retrieval top-K for the CCB; chosen via ablation on evaluation FID (Fig. 7, Table 9a); central to the contrast guidance.
  • Classifier-free guidance scale = 7.5
    Inference hyperparameter (§4.1); hand-chosen, not swept or justified, affects all generated outputs and measured FID.
axioms (5)
  • domain assumption The pre-trained Stable Diffusion v1.5 checkpoint provides a valid generative prior for camouflage scenes and the VQ-VAE latent space supports the decoupling operations.
    CamoDreamer is initialized from SDv1.5 and all modules operate on its latents (§3.1, §4.1); if the base model cannot represent camouflage textures the whole pipeline fails, but this is not independently validated here.
  • domain assumption Cosine similarity in DINOv2 feature space (Eq. 5) retrieves semantically appropriate background anchors for camouflage.
    Representation-driven retrieval is used to build F for every sample; no retrieval-quality ablation isolates it beyond removing anchors (Table 9a), and similar features do not guarantee compatible backgrounds.
  • domain assumption Foreground regions carry high-frequency detail and backgrounds carry low-frequency structure.
    FACB's entire frequency decomposition (§3.5, motivated by [7]) rests on this empirical dichotomy; it is not verified on camouflage data.
  • domain assumption Binary foreground masks are accurate and can enforce separation in cross-attention without leaking.
    Equations 8 and 10 multiply attention outputs by M_fg and 1−M_fg; mask errors would corrupt both streams. Masks are assumed given in LAKE-RED.
  • domain assumption FID/KID against the 5,066 real COD10K images is a valid fidelity measure and reference images are independent of the conditioning/retrieval data.
    §A and §4.1 define the reference set but never document disjointness from the evaluation foregrounds or the training knowledge base; the headline numbers depend on this.

pith-pipeline@v1.3.0-alltime-deepseek · 18200 in / 14362 out tokens · 159347 ms · 2026-08-01T17:03:23.718294+00:00 · methodology

0 comments
read the original abstract

Camouflage image generation (CIG) focuses on generating visually concealed objects that seamlessly blend into their backgrounds. Existing methods typically follow either background-guided paradigms that adapt object appearance via style transfer, or foreground-guided strategies that outpaint surrounding regions conditioned on object features. However, they still suffer from appearance discrepancy and background artifacts. We attribute these limitations to cross-context representation leakage, where object and background cues are entangled in a coupled conditional space, resulting in ambiguous control and degraded camouflage fidelity. To tackle this, we propose a new context-decoupled generative paradigm, termed CamoDreamer, which aims to isolate contextual conditional guidance and explicitly decouple latent camouflage features into coordinated object and background control streams. First, a Contrast-aware Contextual Bridge is designed to model cross-context discrepancies and construct contrast-aware dual conditional guidance. Second, Context-Decoupled Assimilation Streams are employed to separate generative interactions conditioned on the dual guidance, while facilitating background rendering with target-aware cues in the latent space. Finally, a Frequency-Adaptive Contextual Blend module integrates complementary high-frequency textures and low-frequency structures from decoupled features to improve holistic coherence. Extensive experiments demonstrate that CamoDreamer consistently outperforms existing methods with a substantial margin, while maintaining a relatively lightweight design.

Figures

Figures reproduced from arXiv: 2607.17768 by Haoran Chen, Jia Li, Mingcan Ma, Ming Liu, Wenzhuang Wang, Yifan Zhao, Yunlong Che.

Figure 1
Figure 1. Figure 1: Motivation and qualitative comparison between our [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overview of the proposed CamoDreamer framework, which performs CIG via three major steps: (1) a contrast [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Visualization of decoupled camouflage features ( [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Qualitative comparison of camouflage images generated by different CIG methods. The first two columns show real [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: ZoomNet [40] predictions on the COD10K dataset when trained with different synthesized camouflage images. To further evaluate structural alignment, we employ a pretrained COD model (i.e., ZoomNet [40]) to predict object masks and com￾pare them with ground-truth annotations. Higher detection scores (i.e., 𝑆𝑚, 𝐸𝑚, 𝐹𝛽 , and 𝐹 𝑤 𝛽 ) indicate better structural preservation and shape alignment, while lower MAE r… view at source ↗
Figure 6
Figure 6. Figure 6: Visualisations of ablation studies on the proposed [PITH_FULL_IMAGE:figures/full_fig_p008_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Qualitative ablation on the number of background [PITH_FULL_IMAGE:figures/full_fig_p011_7.png] view at source ↗
Figure 9
Figure 9. Figure 9: Qualitative comparison between CamoDreamer and [PITH_FULL_IMAGE:figures/full_fig_p012_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Additional qualitative comparisons of camouflage images generated by different CIG methods. [PITH_FULL_IMAGE:figures/full_fig_p013_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Qualitative visualisations of ablation studies on the proposed components in our CamoDreamer, encompassing [PITH_FULL_IMAGE:figures/full_fig_p014_11.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

81 extracted references · 12 linked inside Pith

  1. [1]

    Radhakrishna Achanta, Sheila Hemami, Francisco Estrada, and Sabine Susstrunk

  2. [2]

    Panagiotis Alimisis, Ioannis Mademlis, Panagiotis Radoglou-Grammatikis, Pana- giotis Sarigiannidis, and Georgios Th Papadopoulos. 2025. Advances in diffusion models for image data augmentation: A review of methods, models, evaluation metrics and future research directions.Artificial Intelligence Review58, 4 (2025), 112

    2025

  3. [3]

    Nan Bao, Yifan Zhao, Wenzhuang Wang, and Jia Li. 2026. Envisioning Beyond the Few: Disentangled Semantics and Primitives for Few-Shot Atypical Layout- to-Image Generation.arXiv preprint arXiv:2605.31266(2026)

    Pith/arXiv arXiv 2026

  4. [4]

    Mikołaj Bińkowski, Danica J Sutherland, Michael Arbel, and Arthur Gretton

  5. [5]

    Chunyuan Chen, Yunuo Cai, Shujuan Li, Weiyun Liang, Bin Wang, and Jing Xu

  6. [6]

    Kai Chen, Enze Xie, Zhe Chen, Yibo Wang, Lanqing Hong, Zhenguo Li, and Dit-Yan Yeung. 2023. Geodiffusion: Text-prompted geometric control for object detection data generation.arXiv preprint arXiv:2306.04607(2023)

    Pith/arXiv arXiv 2023

  7. [7]

    Linwei Chen, Lin Gu, and Ying Fu. 2024. When semantic segmentation meets frequency aliasing.arXiv preprint arXiv:2403.09065(2024)

    Pith/arXiv arXiv 2024

  8. [8]

    Pei-Chi Chen, Yi Yao, Chan-Feng Hsu, HongXia Xie, Hung-Jen Chen, Hong-Han Shuai, and Wen-Huang Cheng. 2025. Foreground focus: Enhancing coherence and fidelity in camouflaged image generation. In2025 IEEE International Conference on Multimedia and Expo (ICME). IEEE, 1–6

    2025

  9. [9]

    Hung-Kuo Chu, Wei-Hsin Hsu, Niloy J Mitra, Daniel Cohen-Or, Tien-Tsin Wong, and Tong-Yee Lee. 2010. Camouflage images.ACM Trans. Graph.29, 4 (2010), 51–1

    2010

  10. [10]

    Biplab Das and Viswanath Gopalakrishnan. 2025. Camouflage anything: Learning to hide using controlled out-painting and representation engineering. InProceed- ings of the Computer Vision and Pattern Recognition Conference. 3603–3613

    2025

  11. [11]

    Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. 2009. Imagenet: A large-scale hierarchical image database. In2009 IEEE conference on computer vision and pattern recognition. Ieee, 248–255

    2009

  12. [12]

    Deng-Ping Fan, Ming-Ming Cheng, Yun Liu, Tao Li, and Ali Borji. 2017. Structure- measure: A new way to evaluate foreground maps. InProceedings of the IEEE international conference on computer vision. 4548–4557

    2017

  13. [13]

    Deng-Ping Fan, Cheng Gong, Yang Cao, Bo Ren, Ming-Ming Cheng, and Ali Borji

  14. [14]

    Deng-Ping Fan, Ge-Peng Ji, Guolei Sun, Ming-Ming Cheng, Jianbing Shen, and Ling Shao. 2020. Camouflaged object detection. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition. 2777–2787

    2020

  15. [15]

    Xiang Gao, Shuai Yang, and Jiaying Liu. 2025. PTDiffusion: Free Lunch for Generating Optical Illusion Hidden Pictures with Phase-Transferred Diffusion Model. InProceedings of the Computer Vision and Pattern Recognition Conference. 18240–18249

    2025

  16. [16]

    Ian J Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. 2014. Generative adversarial nets.Advances in neural information processing systems27 (2014)

    2014

  17. [17]

    Enhanced-alignment measure for binary foreground map evaluation.arXiv preprint arXiv:1805.10421(2018)

    Pith/arXiv arXiv 2018

  18. [18]

    Zhentao He, Changqun Xia, Shengye Qiao, and Jia Li. 2024. Text-prompt camou- flaged instance segmentation with graduated camouflage learning. InProceedings of the 32nd ACM International Conference on Multimedia. 5584–5593

    2024

  19. [19]

    Martin Heusel, Hubert Ramsauer, Thomas Unterthiner, Bernhard Nessler, and Sepp Hochreiter. 2017. Gans trained by a two time-scale update rule converge to a local nash equilibrium.Advances in neural information processing systems30 (2017)

    2017

  20. [20]

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. 2020. Denoising diffusion probabilistic models.Advances in neural information processing systems33 (2020), 6840–6851

    2020

  21. [21]

    Yuchao Gu, Yipin Zhou, Yunfan Ye, Yixin Nie, Licheng Yu, Pingchuan Ma, Kevin Qinghong Lin, and Mike Zheng Shou. 2025. Roictrl: Boosting instance control for visual generation. InProceedings of the Computer Vision and Pattern Recognition Conference. 23658–23667

    2025

  22. [22]

    Xun Huang and Serge Belongie. 2017. Arbitrary style transfer in real-time with adaptive instance normalization. InProceedings of the IEEE international conference on computer vision. 1501–1510

    2017

  23. [23]

    Tero Karras, Samuli Laine, and Timo Aila. 2019. A style-based generator ar- chitecture for generative adversarial networks. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition. 4401–4410

    2019

  24. [24]

    Trung-Nghia Le, Tam V Nguyen, Zhongliang Nie, Minh-Triet Tran, and Aki- hiro Sugimoto. 2019. Anabranch network for camouflaged object segmentation. Computer vision and image understanding184 (2019), 45–56

    2019

  25. [25]

    Duojun Huang, Xinyu Xiong, Jie Ma, Jichang Li, Zequn Jie, Lin Ma, and Guanbin Li. 2024. Alignsam: Aligning segment anything model to open context via reinforcement learning. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition. 3205–3215

    2024

  26. [26]

    Danfeng Li, Hui Zhang, Sheng Wang, Jiacheng Li, and Zuxuan Wu. 2025. Seg2Any: Open-set Segmentation-Mask-to-Image Generation with Precise Shape and Se- mantic Control.arXiv preprint arXiv:2506.00596(2025)

    arXiv 2025

  27. [27]

    Jinyue Li, Yuzhou Yu, Jingjing Yang, Meng Fu, Yani Zhang, Shuyao He, Dian- long Ge, Xin Ning, Yannan Chu, and Qiankun Li. 2026. M3Net: A Macro → Meso→ Micro Clinical-inspired hierarchical 3D Network for Pulmonary Nodule Classification.Information Fusion(2026), 104334

    2026

  28. [28]

    Jiaming Li, Jiacheng Zhang, Jichang Li, Ge Li, Si Liu, Liang Lin, and Guanbin Li

  29. [29]

    Daiqing Li, Huan Ling, Seung Wook Kim, Karsten Kreis, Sanja Fidler, and Antonio Torralba. 2022. Bigdatasetgan: Synthesizing imagenet with pixel-wise annota- tions. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 21330–21340

    2022

  30. [30]

    Yangyang Li, Wei Zhai, Yang Cao, and Zheng-Jun Zha. 2022. Location-free camouflage generation network.IEEE Transactions on Multimedia25 (2022), 5234–5247

    2022

  31. [31]

    Tsung-Yi Lin, Michael Maire, Serge Belongie, James Hays, Pietro Perona, Deva Ramanan, Piotr Dollár, and C Lawrence Zitnick. 2014. Microsoft coco: Common objects in context. InEuropean conference on computer vision. Springer, 740–755

    2014

  32. [32]

    Zhuang Liu, Hanzi Mao, Chao-Yuan Wu, Christoph Feichtenhofer, Trevor Darrell, and Saining Xie. 2022. A convnet for the 2020s. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition. 11976–11986

    2022

  33. [33]

    Andreas Lugmayr, Martin Danelljan, Andres Romero, Fisher Yu, Radu Timofte, and Luc Van Gool. 2022. Repaint: Inpainting using denoising diffusion proba- bilistic models. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition. 11461–11471

    2022

  34. [34]

    Yuheng Li, Haotian Liu, Qingyang Wu, Fangzhou Mu, Jianwei Yang, Jianfeng Gao, Chunyuan Li, and Yong Jae Lee. 2023. Gligen: Open-set grounded text-to-image generation. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition. 22511–22521

    2023

  35. [35]

    Zhengyao Lv, Yuxiang Wei, Wangmeng Zuo, and Kwan-Yee K Wong. 2024. Place: Adaptive layout-semantic fusion for semantic image synthesis. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 9264–9274

    2024

  36. [36]

    Ran Margolin, Lihi Zelnik-Manor, and Ayellet Tal. 2014. How to evaluate fore- ground maps?. InProceedings of the IEEE conference on computer vision and pattern recognition. 248–255

    2014

  37. [37]

    Quang Nguyen, Truong Vu, Anh Tran, and Khoi Nguyen. 2023. Dataset diffusion: Diffusion-based synthetic data generation for pixel-level semantic segmentation. Advances in Neural Information Processing Systems36 (2023), 76872–76892

    2023

  38. [38]

    Alex Nichol, Prafulla Dhariwal, Aditya Ramesh, Pranav Shyam, Pamela Mishkin, Bob McGrew, Ilya Sutskever, and Mark Chen. 2021. Glide: Towards photorealistic image generation and editing with text-guided diffusion models.arXiv preprint arXiv:2112.10741(2021)

    Pith/arXiv arXiv 2021

  39. [39]

    Yunqiu Lv, Jing Zhang, Yuchao Dai, Aixuan Li, Bowen Liu, Nick Barnes, and Deng-Ping Fan. 2021. Simultaneously localize, segment and rank the camouflaged objects. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition. 11591–11601

    2021

  40. [40]

    Youwei Pang, Xiaoqi Zhao, Tian-Zhu Xiang, Lihe Zhang, and Huchuan Lu. 2022. Zoom in and out: A mixed-scale triplet network for camouflaged object detec- tion. InProceedings of the IEEE/CVF Conference on computer vision and pattern recognition. 2160–2170

    2022

  41. [41]

    Bohao Peng, Jian Wang, Yuechen Zhang, Wenbo Li, Ming-Chang Yang, and Jiaya Jia. 2024. Controlnext: Powerful and efficient control for image and video generation.arXiv preprint arXiv:2408.06070(2024)

    Pith/arXiv arXiv 2024

  42. [42]

    Dustin Podell, Zion English, Kyle Lacey, Andreas Blattmann, Tim Dockhorn, Jonas Müller, Joe Penna, and Robin Rombach. 2023. Sdxl: Improving latent diffusion models for high-resolution image synthesis.arXiv preprint arXiv:2307.01952 (2023)

    Pith/arXiv arXiv 2023

  43. [43]

    Yuhang Qian, Haiyan Chen, Wentong Li, Ningzhong Liu, and Jie Qin. 2026. Text- guided Controllable Diffusion for Realistic Camouflage Images Generation. In Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 40. 8529–8537. 9

    2026

  44. [44]

    Maxime Oquab, Timothée Darcet, Théo Moutakanni, Huy Vo, Marc Szafraniec, Vasil Khalidov, Pierre Fernandez, Daniel Haziza, Francisco Massa, Alaaeldin El- Nouby, et al. 2023. Dinov2: Learning robust visual features without supervision. arXiv preprint arXiv:2304.07193(2023)

    Pith/arXiv arXiv 2023

  45. [45]

    Aditya Ramesh, Prafulla Dhariwal, Alex Nichol, Casey Chu, and Mark Chen

  46. [46]

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Björn Ommer. 2022. High-resolution image synthesis with latent diffusion models. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 10684–10695

    2022

  47. [47]

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. 2015. U-net: Convolutional networks for biomedical image segmentation. InInternational Conference on Medical image computing and computer-assisted intervention. Springer, 234–241

    2015

  48. [48]

    Chitwan Saharia, William Chan, Saurabh Saxena, Lala Li, Jay Whang, Emily L Denton, Kamyar Ghasemipour, Raphael Gontijo Lopes, Burcu Karagol Ayan, Tim Salimans, et al. 2022. Photorealistic text-to-image diffusion models with deep language understanding.Advances in neural information processing systems35 (2022), 36479–36494

    2022

  49. [49]

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. 2021. Learning transferable visual models from natural language supervision. InInternational conference on machine learning. PmLR, 8748–8763

    2021

  50. [50]

    Martin Stevens and Sami Merilaita. 2009. Animal camouflage: current issues and new perspectives.Philosophical Transactions of the Royal Society B: Biological Sciences364, 1516 (2009), 423–427

    2009

  51. [51]

    Ming Tao, Hao Tang, Fei Wu, Xiao-Yuan Jing, Bing-Kun Bao, and Changsheng Xu. 2022. Df-gan: A simple and effective baseline for text-to-image synthesis. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 16515–16525

    2022

  52. [52]

    Aaron Van Den Oord, Oriol Vinyals, et al. 2017. Neural discrete representation learning.Advances in neural information processing systems30 (2017)

    2017

  53. [53]

    Lijun Wang, Huchuan Lu, Yifan Wang, Mengyang Feng, Dong Wang, Baocai Yin, and Xiang Ruan. 2017. Learning to detect salient objects with image-level supervision. InProceedings of the IEEE conference on computer vision and pattern recognition. 136–145

    2017

  54. [54]

    Peng Wang, Shuai Bai, Sinan Tan, Shijie Wang, Zhihao Fan, Jinze Bai, Keqin Chen, Xuejing Liu, Jialin Wang, Wenbin Ge, et al. 2024. Qwen2-vl: Enhancing vision-language model’s perception of the world at any resolution.arXiv preprint arXiv:2409.12191(2024)

    Pith/arXiv arXiv 2024

  55. [55]

    Christoph Schuhmann, Romain Beaumont, Richard Vencu, Cade Gordon, Ross Wightman, Mehdi Cherti, Theo Coombes, Aarush Katta, Clayton Mullis, Mitchell Wortsman, et al. 2022. Laion-5b: An open large-scale dataset for training next generation image-text models.Advances in Neural Information Processing Systems 35 (2022), 25278–25294

    2022

  56. [56]

    Weijia Wu, Yuzhong Zhao, Hao Chen, Yuchao Gu, Rui Zhao, Yefei He, Hong Zhou, Mike Zheng Shou, and Chunhua Shen. 2023. Datasetdm: Synthesizing data with perception annotations using diffusion models.Advances in Neural Information Processing Systems36 (2023), 54683–54695

    2023

  57. [57]

    Weijia Wu, Yuzhong Zhao, Mike Zheng Shou, Hong Zhou, and Chunhua Shen

  58. [58]

    Chenxi Xie, Changqun Xia, Tianshu Yu, and Jia Li. 2023. Frequency representation integration for camouflaged object detection. InProceedings of the 31st ACM International Conference on Multimedia. 1789–1797

    2023

  59. [59]

    Jinheng Xie, Yuexiang Li, Yawen Huang, Haozhe Liu, Wentian Zhang, Yefeng Zheng, and Mike Zheng Shou. 2023. Boxdiff: Text-to-image synthesis with training-free box-constrained diffusion. InProceedings of the IEEE/CVF interna- tional conference on computer vision. 7452–7461

    2023

  60. [60]

    Wenjie Xuan, Jing Zhang, Juhua Liu, Bo Du, and Dacheng Tao. 2025. Rethink Sparse Signals for Pose-guided Text-to-image Generation. InProceedings of the IEEE/CVF International Conference on Computer Vision. 15896–15906

    2025

  61. [61]

    Wenzhuang Wang, Yifan Zhao, Mingcan Ma, Ming Liu, Zhonglin Jiang, Yong Chen, and Jia Li. 2025. FICGen: Frequency-Inspired Contextual Disentanglement for Layout-driven Degraded Image Generation. InProceedings of the IEEE/CVF International Conference on Computer Vision. 19097–19107

    2025

  62. [62]

    Zhengyuan Yang, Jianfeng Wang, Zhe Gan, Linjie Li, Kevin Lin, Chenfei Wu, Nan Duan, Zicheng Liu, Ce Liu, Michael Zeng, et al. 2023. Reco: Region-controlled text-to-image generation. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 14246–14255

    2023

  63. [63]

    Han Zhang, Tao Xu, Hongsheng Li, Shaoting Zhang, Xiaogang Wang, Xiaolei Huang, and Dimitris N Metaxas. 2017. Stackgan: Text to photo-realistic image synthesis with stacked generative adversarial networks. InProceedings of the IEEE international conference on computer vision. 5907–5915

    2017

  64. [64]

    Lvmin Zhang, Anyi Rao, and Maneesh Agrawala. 2023. Adding conditional con- trol to text-to-image diffusion models. InProceedings of the IEEE/CVF international conference on computer vision. 3836–3847

    2023

  65. [65]

    Mu Zhang, Yunfan Liu, Yue Liu, Yuzhong Zhao, and Qixiang Ye. 2024. CC- Diff: enhancing contextual coherence in remote sensing image synthesis.arXiv preprint arXiv:2412.08464(2024)

    Pith/arXiv arXiv 2024

  66. [66]

    Qing Zhang, Gelin Yin, Yongwei Nie, and Wei-Shi Zheng. 2020. Deep camouflage images. InProceedings of the AAAI conference on artificial intelligence, Vol. 34. 12845–12852

    2020

  67. [67]

    Shizhou Zhang, Dexuan Kong, Yinghui Xing, Yue Lu, Lingyan Ran, Guoqiang Liang, Hexu Wang, and Yanning Zhang. 2025. Frequency-guided spatial adapta- tion for camouflaged object detection.IEEE Transactions on Multimedia27 (2025), 72–83

    2025

  68. [68]

    Changzhi Yang, Huihui Pan, Jue Wang, and Yuanduo Hong. 2025. TrajDiff: Trajectory Prediction With Diffusion Probabilistic Models.IEEE Transactions on Image Processing34 (2025), 8257–8270

    2025

  69. [69]

    Yuxuan Zhang, Huan Ling, Jun Gao, Kangxue Yin, Jean-Francois Lafleche, Adela Barriuso, Antonio Torralba, and Sanja Fidler. 2021. Datasetgan: Efficient labeled data factory with minimal human effort. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 10145–10155

    2021

  70. [70]

    Peiang Zhao, Han Li, Ruiyang Jin, and S Kevin Zhou. 2025. LoCo: Training-Free Layout-to-Image Synthesis with Localized Constraints. InProceedings of the 33rd ACM International Conference on Multimedia. 9481–9490

    2025

  71. [71]

    Pancheng Zhao, Peng Xu, Pengda Qin, Deng-Ping Fan, Zhicheng Zhang, Guoli Jia, Bowen Zhou, and Jufeng Yang. 2024. Lake-red: Camouflaged images generation by latent background knowledge retrieval-augmented diffusion. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 4092–4101

    2024

  72. [72]

    Chuanxia Zheng, Tat-Jen Cham, Jianfei Cai, and Dinh Phung. 2022. Bridging global context interactions for high-fidelity image completion. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition. 11512–11522

    2022

  73. [73]

    Dewei Zhou, You Li, Fan Ma, Xiaoting Zhang, and Yi Yang. 2024. Migc: Multi- instance generation controller for text-to-image synthesis. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 6818–6828

    2024

  74. [74]

    Camouflaged fish in an underwater rocky-sandy habitat

    Minfeng Zhu, Pingbo Pan, Wei Chen, and Yi Yang. 2019. Dm-gan: Dynamic mem- ory generative adversarial networks for text-to-image synthesis. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition. 5802–5810. 10 Supplementary Materials This supplementary material provides additional experimental de- tails and analyses omitted fr...

    2019

  75. [75]

    Xinlong Zhang, Zejian Li, Wei Li, Xiaoyu Zhang, Jia Wei, Chengyu Lin, and Yongchuan Tang. 2025. ObjCtrl: Object-based Control Relaxation for Conditional Text-to-Image Generation. InProceedings of the 33rd ACM International Conference on Multimedia. 10064–10073

    2025

  76. [2009]

    In2009 IEEE conference on computer vision and pattern recognition

    Frequency-tuned salient region detection. In2009 IEEE conference on computer vision and pattern recognition. IEEE, 1597–1604

  77. [2018]

    Demystifying mmd gans.arXiv preprint arXiv:1801.01401(2018)

    Pith/arXiv arXiv 2018

  78. [2022]

    Hierarchical text-conditional image generation with clip latents.arXiv preprint arXiv:2204.061251, 2 (2022), 3

    Pith/arXiv arXiv 2022

  79. [2023]

    InProceedings of the IEEE/CVF International Conference on Computer Vision

    Diffumask: Synthesizing images with pixel-level annotations for semantic segmentation using diffusion models. InProceedings of the IEEE/CVF International Conference on Computer Vision. 1206–1217

  80. [2024]

    InProceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Learning background prompts to discover implicit knowledge for open vocabulary object detection. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition. 16678–16687

Showing first 80 references.