Pith. sign in

REVIEW 4 major objections 4 minor 46 references

DAC-Pose: Dual-Agent Collaborative Framework for Pose-Guided Human Generation

T0 review · 4 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read DAC-Pose claims that a collaborative dual-agent system—one agent reasoning in language about unseen body regions, the other encoding pose-induced spatial discrepancies—improves pose-guided human image generation and beats all compared…

desk verdict The DAVE agent's use of the target image at inference is either test-time leakage or an unspecified gap, and the thin metric gains do not compensate. read the letter →

arxiv 2608.04622 v1 pith:A6WOVQBZ submitted 2026-08-05 cs.CV

Haotian Yang , Zhile Yang , Huiyu Zhou , Xin Sun This is my paper · ORCID
classification cs.CV
keywords pose-guidedpersonimagegenerationAIagentssemanticreasoningdiscrepancy-awarevisualencodinglatentdiffusioncross-viewposeinversionout-of-viewbodycompletionDeepFashion
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper claims that pose-guided human image generation can be improved by replacing a single passive pipeline with a collaborative pair of agents: one that reasons in language about body regions invisible in the source view, and one that explicitly measures the spatial mismatch between source and target poses. It says this dual-agent design lets a diffusion model synthesize clothing patterns, identity features, and newly exposed body parts under drastic viewpoint changes instead of hallucinating them. On DeepFashion and Market-1501, the reported numbers show DAC-Pose ahead of all fourteen comparison methods on SSIM, LPIPS, and FID. The reason this matters is that virtual try-on, digital avatars, and film production depend on preserving appearance when the pose changes radically.

What carries the argument

The central machinery is the collaborative dual-agent loop. The PSR agent converts pose pairs into spatial semantic constraints (a yaw-shift score derived from keypoint distances around a nose-neck reference axis, plus an LLM-detected extended body region) and then into three MLLM-generated text descriptions fused by reliability analysis into a conditioning embedding. The DAVE agent computes a discrepancy encoding $c = \text{CA}(a,b,b)$ from source and target VAE embeddings and fuses it as $g = \text{CA}(c,d,d)$ to steer the denoising UNet. The paper's argument is that semantic deduction supplies what the source view does not show, while discrepancy encoding supplies where the two views differ, and that both are needed.

What would settle it

Run the released DAC-Pose inference code on a hold-out DeepFashion split and check whether the VAE encoder ever receives the ground-truth target image $I_t$: if it does, the reported gains may measure target leakage rather than agent reasoning. A cleaner test replaces the target embedding in Eq. (14) with a prediction estimated from the source image and target pose; if SSIM and FID fall to roughly the "without DAVE" ablation levels, the discrepancy encoding is not performing the claimed work.

Watch

Extended reading notes

Core claim

DAC-Pose reformulates single-view human generation as a dual-agent system. The Prior Semantic Reasoning agent uses DWPose skeletons to estimate yaw shift and detect expanded body parts, then has a multimodal LLM produce three text descriptions (source appearance, extended region, perspective difference) that are merged by reliability analysis into a semantic embedding. The Discrepancy-Aware Visual Encoding agent encodes source and target images with a frozen VAE and computes a discrepancy encoding via cross-attention, then fuses it with the reference-branch feature to condition the denoising UNet. With these two conditions, the paper reports the best SSIM, LPIPS, and FID on DeepFashion and Market-1501, and ablations show removing either agent degrades the numbers, with DAVE removal causing the larger drop.

Load-bearing premise

The DAVE agent's discrepancy encoding is computed from the target image $I_t$, but at inference the target image is the unknown output, and the paper does not say how the target embedding is obtained at test time.

Editorial extensions

If this is right

  • DAC-Pose reports the best SSIM, LPIPS, and FID on both DeepFashion and Market-1501, so an agent-based conditioning design can compete with and slightly surpass strong diffusion baselines.
  • Removing either agent lowers measured performance, which implies the semantic priors and discrepancy constraints are complementary rather than redundant.
  • Because DAVE removal hurts more than PSR removal, spatial discrepancy encoding carries a larger share of the visual-fidelity improvement under the paper's settings.
  • The framework generalizes across resolutions and environments, since gains appear on 256×176 DeepFashion and 128×64 Market-1501.

Reading between the lines

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

  • If the inference-time DAVE gap is closed by predicting the target embedding from the source image and target pose, the reported margin over IMAGPose is likely to shrink; the paper's current formulation leaves this unresolved.
  • The PSR agent's quality depends on the multimodal LLM's descriptions, so a test that feeds oracle or ground-truth captions could separate the benefit of textual priors from the benefit of the fusion machinery.
  • The same dual-agent pattern could transfer to video human animation or virtual try-on, where large viewpoint changes are the dominant failure mode, but this is an extension the paper does not demonstrate.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 4 minor

Summary. This manuscript proposes DAC-Pose, a dual-agent framework for pose-guided human image generation. The Prior Semantic Reasoning (PSR) agent uses an MLLM to generate textual semantic priors for unseen or view-shifted regions, while the Discrepancy-Aware Visual Encoding (DAVE) agent computes a discrepancy encoding between source and target embeddings and fuses it into a denoising diffusion U-Net. The authors report state-of-the-art SSIM, LPIPS, and FID on DeepFashion and Market-1501, along with ablations showing that both agents contribute. I read the paper with particular attention to the test-time protocol for the DAVE agent, since the discrepancy encoding appears to be computed from the target image.

Significance. The agent-based framing, especially the combination of explicit spatial constraints and textual semantic priors, is timely and could be useful for extreme pose transfer. The paper also promises code release, which is a positive feature for reproducibility. However, the central quantitative claim is not supported as written because the DAVE agent's discrepancy encoding is computed from the target image, and the reported gains over the strongest baseline are extremely small and lack any uncertainty quantification. If the target-image dependence were resolved and the evaluation repeated with proper statistical testing, the framework could still be of interest, but the current evidence does not establish state-of-the-art performance.

major comments (4)
  1. [Section 3.3.1, Eqs. (14)-(16)] The target image It enters the DAVE pipeline as b = E_vae(It) and is used to form c = CA(a, b, b) and then g = CA(c, d, d). At inference time, It is precisely the unknown image that the model is supposed to produce, yet the paper does not state that Eq. (14) is training-only, nor does it specify how b or c is obtained at test time (for example, from the target pose, from an intermediate generation, or from a separate estimator). The Section 4.2 results and Table 2 ablations are reported without any such caveat, so the reported metrics are consistent with the model being conditioned on the ground-truth target. This is a load-bearing gap: either the evaluation leaks information from the ground truth, or a critical part of the test-time architecture is unspecified. The authors must clarify the exact training and inference protocols and, if necessary, re-evaluate the method without target-image information.
  2. [Section 4.2, Table 1] The reported advantage over the strongest prior work is numerically very small: on DeepFashion, SSIM is 0.7572 versus 0.7561, LPIPS is 0.1274 versus 0.1284, and FID is 5.8547 versus 5.8738. No error bars, confidence intervals, or repeated-run statistics are provided, and no significance test is performed. Because diffusion sampling is stochastic and the pipeline includes LLM-based description generation, these differences are within plausible run-to-run variation. The headline claim of consistent superiority therefore needs a statistical foundation before it can be accepted.
  3. [Section 3.2.1, Eq. (1)] The target pose map is defined as Pt = Epose(It), using the target image itself. If the target pose is part of the task input, the authors should say so explicitly and distinguish it from any later use of It in DAVE. If the target pose is derived from the ground-truth target image during evaluation, that is standard for this benchmark but should be stated clearly. The current wording blurs the boundary between the provided pose condition and the unknown target appearance, compounding the concern raised in the first major comment.
  4. [Section 3.2.1, Eq. (4), and Section 3.2.2, Eq. (12)] Several components are left unspecified. The viewpoint keypoint weights omega_m are described only as satisfying a sum-to-one constraint; the quantization function Q_view in Eq. (6) is not defined; and the reliability analysis function F_rel in Eq. (12) has no description at all. These are not cosmetic details: they determine the spatial semantic constraint C_spa that drives the PSR agent. Without specifying them, the method cannot be reproduced from the manuscript, and the paper's reproducibility claims are weakened.
minor comments (4)
  1. [Throughout] The term 'DA VE' is typeset with an unusual space in many places; please use 'DAVE' consistently.
  2. [Section 2.1] The sentence beginning 'Despite their notable success under mild pose variations...' appears twice in near-identical form at the end of the subsection; please remove the duplicate.
  3. [Figure 2] Figure 2 shows a 'Target Image' input and a 'Source Reconstruction Loss' that are not explained in the main text; these should be explicitly reconciled with the training and inference descriptions.
  4. [Section 4.1] The paper names GPT-5.5 as the multimodal foundation model but gives no version, decoding parameters, or prompting scheme; please provide these details or a reference to a reproducible configuration.

Circularity Check

1 steps flagged · score 8.0 of 10

DAVE's discrepancy encoding is computed from the target image It, making the reported test-set generation circular by construction.

  1. self definitional [Section 3.3.1, Eqs. (14)-(15); Section 3.3.2, Eqs. (16)-(17); Section 4.2.1, Table 1]
    "Specifically, the source image Is and target image It are encoded via a VAE encoder [9]: a = Evae(Is), b = Evae(It), (14) ... c = CA(a, b, b), (15) ... CDAVE = g. (17) CDAVE is used in the second-stage generation as a discrepancy-aware condition, enabling the denoising network to emphasize regions with large source-target discrepancies."

    At inference, It is exactly the image that DAC-Pose is supposed to synthesize, yet Eq. (14) feeds It into the VAE to produce b, Eq. (15) builds the 'discrepancy encoding' c from b, and Eqs. (16)-(17) inject C_DAVE into the denoising fusion. The only limiting phrase is that Sec. 3.3.1 registers correspondences 'during training'; the paper never states that Eq. (14) is bypassed at test time, never provides a substitute target embedding (e.g., from the target pose or from an intermediate sample), and Sec. 4.2 reports final test-set numbers using this condition. Therefore the reported SSIM/LPIPS/FID gains are not predictions of It from Is and Pt; they are outputs conditioned on It itself.

full rationale

The only load-bearing circular step is the DAVE agent's dependence on the target image It. The PSR agent and text-semantic branches condition only on the source image and the target pose, and the DWPose pose extractor is an external tool, so those components are not circular. However, the flagship quantitative claim in Table 1 depends on C_DAVE, which is built from b = Evae(It). If this is truly used at test time, the model is given the ground truth target as a conditioning input, making the evaluation circular; if it is training-only, the paper leaves the test-time DAVE mechanism unspecified. Either way, the reported state-of-the-art numbers do not support the claimed agent-driven paradigm as written. This warrants a score of 8 rather than a lower score, because the central result reduces to target information by the paper's own equations.

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

The central claim rests on the correctness of the LLM's semantic inference and on the availability of the target image for the DAVE module. The paper introduces no new physical or mathematical entities, but it relies on several unspecified hand-chosen parameters and unverified assumptions about the LLM and pose estimator.

free parameters (4)
  • viewpoint keypoint weights omega_m = not specified
    In Eq. 4, the weights omega_m sum to 1 but their values are never given; they are hand-chosen or tuned without reporting.
  • viewpoint quantization thresholds = not specified
    The quantization function Qview in Eq. 6 discretizes the viewpoint score into states but the thresholds are not defined.
  • GPT-5.5 generation parameters = not specified
    Temperature, top-p, prompt templates, and model version for the multimodal LLM are not reported, making the text priors non-deterministic and unreproducible.
  • reliability analysis function Frel = not specified
    The function Frel in Eq. 12 that aggregates the three description branches is not defined; it is a hand-crafted or learned module with no specification.
assumptions (4)
  • domain assumption The multimodal LLM can correctly infer appearance attributes of unseen body regions from the source image and spatial semantic constraints.
    The PSR agent's entire contribution relies on the LLM making plausible guesses about occluded regions such as back-view clothing or extended limbs; the paper provides no systematic verification of inference accuracy.
  • domain assumption DWPose provides sufficiently accurate 2D skeletons for source and target poses.
    Eq. 1 uses DWPose to extract keypoints; if the pose estimator fails under extreme views, the downstream yaw-shift estimation and extended part detection would be incorrect.
  • domain assumption SSIM, LPIPS, and FID capture the relevant perceptual quality improvements.
    These metrics are used as the sole evidence of superiority; the paper does not include user studies or other task-specific measures.
  • domain assumption Stable Diffusion, Sentence-BERT, and the pretrained PoseGuider provide adequate backbones for the task.
    The method inherits all assumptions of these pretrained models, including their biases and failure modes, without re-validation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of DAC-Pose: Dual-Agent Collaborative Framework for Pose-Guided Human Generation." pith.science (2026). https://pith.science/paper/A6WOVQBZ

@misc{pith2026260804622,
  author       = {Pith},
  title        = {Pith review of: DAC-Pose: Dual-Agent Collaborative Framework for Pose-Guided Human Generation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/A6WOVQBZ}},
  note         = {Machine review of arXiv:2608.04622}
}
read the original abstract

AI agents have emerged as a powerful new paradigm in generative image synthesis, enabling systems to perform complex semantic reasoning rather than passive pixel-level mapping. In pose-guided human generation, conventional methods inevitably produce severe visual artifacts under drastic viewpoint shifts, fundamentally because they lack the cognitive capacity to logically deduce unseen regions and model complex spatial deformations. To bridge this gap, we propose DAC-Pose, a novel agent-driven multimodal framework that reformulates single-view human generation as a collaborative dual-agent system. DAC-Pose integrates two complementary components, namely, the Prior Semantic Reasoning (PSR) agent and the Discrepancy-Aware Visual Encoding (DAVE) agent. Functioning as a cognitive engine, PSR utilizes collaborative reasoning to deduce the fine-grained attributes of unseen regions. Concurrently, acting as a specialized visual perception agent, DAVE quantifies and encodes viewpoint-induced spatial misalignments, continuously feeding robust spatial constraints back into the generative process. This autonomous feedback loop between semantic deduction and visual perception ensures high-fidelity detail synthesis. Extensive experiments on the DeepFashion and Market-1501 benchmarks validate the superiority of our agent-driven paradigm. Notably, DAC-Pose excels in preserving texture alignment and identity consistency under drastic viewpoint changes. The code is available at https://github.com/AIVRC/DAC-Pose.

Figures

Figures reproduced from arXiv: 2608.04622 by the authors.

Figure 1
Figure 1. Comparisons across two challenging cases. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. The overall architecture of DAC-Pose framework with (a) Prior Semantic Reasoning Agent and [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Qualitative comparison under cross-view pose inversion on DeepFashion. [PITH_FULL_IMAGE:figures/full_fig_p017_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Qualitative comparison under out-of-view body completion on DeepFashion. [PITH_FULL_IMAGE:figures/full_fig_p017_4.png]
Figure 5
Figure 5. Figure 5: Qualitative ablation under (a) cross-view pose inversion and (b) out-of-view body completion. [PITH_FULL_IMAGE:figures/full_fig_p020_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

46 extracted references · 44 canonical work pages

  1. [1]

    L. Ma, X. Jia, Q. Sun, B. Schiele, T. Tuytelaars, L. Van Gool, Pose guided per- son image generation, in: Advances in Neural Information Processing Systems, V ol. 30, 2017

    work page 2017

  2. [2]

    Zhang, L

    P. Zhang, L. Yang, J.-H. Lai, X. Xie, Exploring dual-task correlation for pose- guided person image generation, in: Proceedings of the IEEE /CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 7703–7712

    work page 2022

  3. [3]

    Z. Zhu, T. Huang, B. Shi, M. Yu, B. Wang, X. Bai, Appearance and pose-guided human generation: A survey, ACM Computing Surveys 56 (5) (2024) 1–35

    work page 2024

  4. [4]

    F. Shen, H. Ye, J. Zhang, C. Wang, X. Han, W. Yang, Advancing pose-guided image synthesis with progressive conditional di ffusion models, in: The Twelfth International Conference on Learning Representations, 2024

    work page 2024

  5. [5]

    Y . Lu, M. Zhang, A. J. Ma, X. Xie, J.-H. Lai, Coarse-to-fine latent di ffusion for pose-guided person image synthesis, in: Proceedings of the IEEE /CVF Confer- ence on Computer Vision and Pattern Recognition, 2024, pp. 6420–6429

    work page 2024

  6. [6]

    J. Liu, J. Zhang, P. Rota, N. Sebe, Multi-focal conditioned latent diffusion for per- son image synthesis, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2025, pp. 16019–16028

    work page 2025

  7. [7]

    D. P. Kingma, M. Welling, Auto-encoding variational bayes, in: International Conference on Learning Representations, 2014

    work page 2014

  8. [8]

    Radford, J

    A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sastry, A. Askell, P. Mishkin, J. Clark, G. Krueger, I. Sutskever, Learning transferable visual models from natural language supervision, in: Proceedings of the 38th In- ternational Conference on Machine Learning, V ol. 139 of Proceedings of Machine Learning Research, PMLR, 2021, pp. 8748–8763

    work page 2021

Show all 46 references
  1. [9]

    Rombach, A

    R. Rombach, A. Blattmann, D. Lorenz, P. Esser, B. Ommer, High-resolution im- age synthesis with latent diffusion models, in: Proceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, 2022, pp. 10674–10685. 22

    2022

  2. [10]

    D. Feng, P. Guo, E. Peng, M. Zhu, W. Yu, P. Wang, Posellava: Pose centric multimodal llm for fine-grained 3d pose manipulation, Proceedings of the AAAI Conference on Artificial Intelligence 39 (3) (2025) 2951–2959

    2025

  3. [11]

    X. Han, X. Zhu, J. Deng, Y .-Z. Song, T. Xiang, Controllable person image syn- thesis with pose-constrained latent di ffusion, in: Proceedings of the IEEE /CVF International Conference on Computer Vision, 2023, pp. 22711–22720

    2023

  4. [12]

    F. Shen, J. Tang, IMAGPose: A unified conditional framework for pose-guided person generation, in: Advances in Neural Information Processing Systems, V ol. 37, 2024, pp. 6246–6266

    2024

  5. [13]

    A. K. Bhunia, S. Khan, H. Cholakkal, R. M. Anwer, J. Laaksonen, M. Shah, F. S. Khan, Person image synthesis via denoising di ffusion model, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2023, pp. 5968–5976

    2023

  6. [14]

    I. J. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, Y . Bengio, Generative adversarial nets, Advances in neural infor- mation processing systems 27 (2014)

    2014

  7. [15]

    Loss functions for pose guided person image generation, Pattern Recognition 122 (2022) 108351

    2022

  8. [16]

    Siarohin, E

    A. Siarohin, E. Sangineto, S. Lathuili‘ere, N. Sebe, Deformable GANs for pose- based human image generation, in: Proceedings of the IEEE Conference on Com- puter Vision and Pattern Recognition, 2018, pp. 3408–3416

    2018

  9. [17]

    Esser, E

    P. Esser, E. Sutter, B. Ommer, A variational u-net for conditional appearance and shape generation, in: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2018

    2018

  10. [18]

    Y . Ren, X. Yu, J. Chen, T. H. Li, G. Li, Deep image spatial transformation for person image generation, in: Proceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition, 2020, pp. 7687–7696. 23

    2020

  11. [19]

    Exploiting appearance transfer and multi-scale context for e fficient person image generation, Pattern Recognition 124 (2022) 108451

    2022

  12. [20]

    Fatnet: Feature-alignment transformer network for human pose transfer, Pattern Recognition 165 (2025) 111626

    2025

  13. [21]

    X. Han, X. Hu, W. Huang, M. R. Scott, ClothFlow: A flow-based model for clothed person generation, in: Proceedings of the IEEE /CVF International Con- ference on Computer Vision, 2019, pp. 10470–10479

    2019

  14. [22]

    Pose-driven attention-guided image generation for person re-identification, Pat- tern Recognition 137 (2023) 109246

    2023

  15. [23]

    One-shot novel view and pose human image synthesis via 3d prior guided di ffu- sion model, Pattern Recognition 179 (2026) 113644

    2026

  16. [24]

    T. Wang, L. Li, K. Lin, Y . Zhai, C.-C. Lin, Z. Yang, H. Zhang, Z. Liu, L. Wang, DisCo: Disentangled control for realistic human dance generation, in: Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 9326–9336

    2024

  17. [25]

    Brown, B

    T. Brown, B. Mann, N. Ryder, M. Subbiah, J. D. Kaplan, P. Dhariwal, A. Nee- lakantan, P. Shyam, G. Sastry, A. Askell, et al., Language models are few-shot learners, Advances in neural information processing systems 33 (2020) 1877– 1901

    2020

  18. [26]

    D. Fan, T. Chen, M. Wang, R. Ma, Q. Tang, Z. Yi, Q. Wang, L. Chang, One-shot learning for pose-guided person image synthesis in the wild, in: ICASSP 2025- 2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), IEEE, 2025, pp. 1–5

    2025

  19. [27]

    T. Lee, D. Lee, M. Kang, Pointt2i: Llm-based text-to-image generation via key- points, Neurocomputing 668 (c) (2026) 132363

    2026

  20. [28]

    L. Hu, X. Gao, P. Zhang, K. Sun, B. Zhang, L. Bo, Animate anyone: Consistent and controllable image-to-video synthesis for character animation, in: Proceed- 24 ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 8153–8163

    2024

  21. [29]

    Z. Xu, J. Zhang, J. H. Liew, H. Yan, J.-W. Liu, C. Zhang, J. Feng, M. Z. Shou, MagicAnimate: Temporally consistent human image animation using di ffusion model, in: Proceedings of the IEEE /CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2024, pp. 1481–1490

    2024

  22. [30]

    H. Liu, C. Li, Q. Wu, Y . J. Lee, Visual instruction tuning, in: Advances in Neural Information Processing Systems, V ol. 36, 2023, pp. 34892–34916

    2023

  23. [31]

    X. Ju, A. Zeng, C. Zhao, J. Wang, L. Zhang, Q. Xu, HumanSD: A native skeleton-guided diffusion model for human image generation, in: Proceedings of the IEEE /CVF International Conference on Computer Vision (ICCV), 2023, pp. 15988–15998

    2023

  24. [32]

    Z. Yang, A. Zeng, C. Yuan, Y . Li, Effective whole-body pose estimation with two- stages distillation, in: Proceedings of the IEEE/CVF International Conference on Computer Vision Workshops, 2023, pp. 4212–4222

    2023

  25. [33]

    Z. Liu, P. Luo, S. Qiu, X. Wang, X. Tang, DeepFashion: Powering robust clothes recognition and retrieval with rich annotations, in: Proceedings of the IEEE Con- ference on Computer Vision and Pattern Recognition (CVPR), 2016, pp. 1096– 1104

    2016

  26. [34]

    Zheng, L

    L. Zheng, L. Shen, L. Tian, S. Wang, J. Wang, Q. Tian, Scalable person re- identification: A benchmark, in: Proceedings of the IEEE International Confer- ence on Computer Vision, 2015, pp. 1116–1124

    2015

  27. [35]

    Reimers, I

    N. Reimers, I. Gurevych, Sentence-BERT: Sentence embeddings using siamese BERT-networks, in: Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP), Associa...

    2019

  28. [36]

    D. P. Kingma, J. Ba, Adam: A method for stochastic optimization, in: Interna- tional Conference on Learning Representations, 2015

    2015

  29. [37]

    J. Song, C. Meng, S. Ermon, Denoising di ffusion implicit models, in: Interna- tional Conference on Learning Representations, 2021

    2021

  30. [38]

    Z. Wang, A. C. Bovik, H. R. Sheikh, E. P. Simoncelli, Image quality assessment: From error visibility to structural similarity, IEEE Transactions on Image Pro- cessing 13 (4) (2004) 600–612

    2004

  31. [39]

    Zhang, P

    R. Zhang, P. Isola, A. A. Efros, E. Shechtman, O. Wang, The unreasonable ef- fectiveness of deep features as a perceptual metric, in: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2018, pp. 586–595

    2018

  32. [40]

    Heusel, H

    M. Heusel, H. Ramsauer, T. Unterthiner, B. Nessler, S. Hochreiter, GANs trained by a two time-scale update rule converge to a local nash equilibrium, in: Advances in Neural Information Processing Systems, V ol. 30, 2017, pp. 6626–6637

    2017

  33. [41]

    Z. Zhu, T. Huang, B. Shi, M. Yu, B. Wang, X. Bai, Progressive pose attention transfer for person image generation, in: Proceedings of the IEEE /CVF Confer- ence on Computer Vision and Pattern Recognition, 2019, pp. 2342–2351

    2019

  34. [42]

    Y . Men, Y . Mao, Y . Jiang, W.-Y . Ma, Z. Lian, Controllable person image synthesis with attribute-decomposed GAN, in: Proceedings of the IEEE /CVF Conference on Computer Vision and Pattern Recognition, 2020, pp. 5083–5092

    2020

  35. [43]

    Zhang, K

    J. Zhang, K. Li, Y .-K. Lai, J. Yang, PISE: Person image synthesis and editing with decoupled GAN, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021, pp. 7978–7986

    2021

  36. [44]

    Y . Ren, X. Fan, G. Li, S. Liu, T. H. Li, Neural texture extraction and distribution for controllable person image synthesis, in: Proceedings of the IEEE /CVF Con- ference on Computer Vision and Pattern Recognition, 2022, pp. 13525–13534

    2022

  37. [45]

    X. Zhou, M. Yin, X. Chen, L. Sun, C. Gao, Q. Li, Cross attention based style distribution for controllable person image synthesis, in: European conference on computer vision, Springer, 2022, pp. 161–178. 26

    2022

  38. [46]

    Shang, H

    Z. Shang, H. Liu, R. Zhang, Z. Wei, T. Feng, PMMD: A pose-guided multi-view multi-modal diffusion for person generation, in: IEEE International Conference on Acoustics, Speech and Signal Processing, 2026, pp. 4321–4325. 27

    2026

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.