Pith. sign in

REVIEW 4 major objections 6 minor 1 cited by

Controlling the Latent Diffusion Model for Generative Image Shadow Removal via Residual Generation

T0 review · 4 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

Pith's one-line read A diffusion model that removes shadows by generating only the shadow residual.

desk verdict A solid residual-generation adaptation of latent diffusion for shadow removal with strong reported numbers, but the inference text has a sign error and no code is released, so the numbers aren't yet reproducible. read the letter →

arxiv 2412.02322 v1 pith:Y4NRPVXO submitted 2024-12-03 cs.CV

classification cs.CV
keywords shadowremovallatentdiffusionmodelresidualgenerationimagerestorationsamplingdetailpreservationgenerativepriorControlNet
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

This paper argues that the best way to apply a large pre-trained diffusion model to shadow removal is to have it generate and refine the shadow residual—the difference between the shadowed and shadow-free image—rather than regenerate the shadow-free image from scratch. That shift lets the model lean on the existing image's details, so the content outside and inside the shadow is preserved instead of being reinvented. To make this work, the paper adds a residual schedule to the diffusion forward process, a noise–residual decomposition to the sampling step, a cross-timestep training strategy that uses the model itself to correct its own trajectory, and a detail-preserving decoder that keeps high-frequency information intact. On ISTD+ and SRD, the method reports the best LPIPS and FID scores among the compared mask-free methods, and the highest PSNR (33.38 dB on ISTD+). The core claim is that a frozen generative backbone can be turned into a high-fidelity deterministic restorer if the diffusion process is steered to edit only what needs changing.

What carries the argument

The key machinery is the residual generation diffusion process: a shadow residual schedule $\bar{\beta}_t$ is layered on top of the unchanged DDIM noise schedule, so the forward latent becomes $z_t = \sqrt{\alpha_t}(z_0 + \bar{\beta}_t r) + \sqrt{1-\alpha_t}\epsilon$ with $r = z_s - z_0$. During sampling, a Noise-Residual Decomposition (NRD) step extracts the estimated residual $\hat{r}_t = z_s - \hat{z}_0^t$ and subtracts the residual's contribution from the network output to recover the pure noise estimate (Eq. 12), then recombines them with the shadow latent to produce the next latent. The residual term is what lets the frozen pre-trained diffusion backbone estimate both noise and shadow correction from one output, while a ControlNet branch (initialized from the pre-trained model) learns the residual. Around this, the cross-timestep self-enhancement training uses an EMA weight copy to produce pseudo-inputs for the main network, and the detail-preserving decoder adds a controller with zero-initialized deformable convolutions that inject multi-scale shadow-image features into the decoding process and adds an image-domain residual.

What would settle it

Run the trained model on shadow-free test images: if the residual schedule and decomposition are correct, the output should reconstruct the input almost exactly (a near-zero residual). Any invented texture or brightness change in those inputs would show that the model is generating content rather than only removing shadows.

Watch

Extended reading notes

Core claim

The central claim is that shadow removal can be reformulated as residual generation within a pre-trained latent diffusion model, yielding higher fidelity than both full regeneration and prior residual-based diffusion methods. The paper introduces a forward process in which the noisy latent interpolates between the shadowed and shadow-free latents via a residual schedule, and a sampling rule that decomposes the network output into a noise component and a residual component at every step. Because the backbone network is frozen and only a ControlNet-style branch is trained, the generative prior from the large model is retained while the new branch learns the shadow-specific correction. The authors further claim that a cross-timestep self-enhancement strategy—where an EMA copy of the network generates pseudo-inputs at earlier timesteps—prevents error accumulation, and that a decoder equipped with a controller, multi-scale skip connections, and zero-initialized deformable convolutions preserves details the original VQ-GAN decoder would distort. The evidence is the reported state-of-the-art perceptual metrics and the ablation results.

Load-bearing premise

The load-bearing premise is that the pre-trained diffusion network's learned denoising behavior remains valid when its input latent is a mix of Gaussian noise and a shadow residual, rather than pure noised images.

Editorial extensions

If this is right

  • Mask-free shadow removal can reach higher perceptual fidelity than previous specialized methods, as measured by LPIPS and FID on ISTD+ and SRD.
  • A frozen pre-trained diffusion backbone can be reused for a deterministic restoration task by only training a control branch, avoiding full retraining.
  • The cross-timestep self-enhancement strategy reduces error accumulation, which should allow the model to use fewer sampling steps without degradation.
  • The detail-preserving decoder can be applied to any latent diffusion pipeline where the VQ-GAN codec destroys high-frequency content, not just shadow removal.
  • With a mask provided, the same framework improves further (PSNR 34.73, LPIPS 0.0228 on ISTD+), suggesting the architecture scales with task information.

Reading between the lines

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

  • The residual-generation trick likely generalizes to other deterministic image-to-image tasks, such as dehazing, deraining, or low-light enhancement, wherever the input already contains most of the output content.
  • The noise-residual decomposition might be a general recipe for controlled editing with diffusion models: any editable attribute could be encoded as a residual schedule on top of a frozen backbone.
  • Since the backbone is frozen, swapping in different control branches could turn a single large model into a multi-task restorer, with only the small branch retrained per task.
  • A testable extension is to measure whether non-shadow regions pass through the pipeline unchanged; if the skip connections are the main fidelity driver, then removing the residual schedule should only slightly hurt non-shadow PSNR.
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 / 6 minor

Summary. The paper proposes a latent-diffusion shadow-removal framework that generates and refines image residuals instead of regenerating shadow-free images from scratch. It keeps a pretrained Paint-by-Example inpainting backbone, adds a ControlNet-style branch, introduces a shadow residual schedule into the forward diffusion process, trains with an EMA-based cross-timestep self-enhancement scheme, and augments the VQGAN decoder with a control branch and multi-scale skip connections. Experiments on ISTD+ and SRD report improved LPIPS and FID scores over prior mask-free methods, competitive PSNR/SSIM, and an extension to mask-available shadow removal.

Significance. If the described algorithm is reproducible as written, the paper makes a useful contribution: the residual-generation formulation is a principled way to retain the generative prior of a large pretrained diffusion model while preserving input fidelity, and the cross-timestep self-enhancement strategy plus the detail-preserving decoder are supported by the ablation study. The reported LPIPS/FID gains on ISTD+ and SRD are substantial, and the mask-available variant is competitive with current state of the art. However, the manuscript as written contains an internal sign inconsistency in the final sampling reconstruction, inconsistent noise-schedule notation, and no released code, so the empirical claims cannot currently be reproduced from the text. These issues are fixable but require a careful revision.

major comments (4)
  1. [Section III-A, Eq. (10)-(11) and final paragraph] There is a sign inconsistency in the final reconstruction. Equation (10) defines the estimated residual as r_hat_t = z_s - z_hat_t^0. Substituting this into Eq. (11) at the final step with beta_bar_0 = 0 gives z'_0 = z_s + (beta_bar_0 - 1) r_hat_1 = z_hat_1^0, which is the predicted shadow-free latent. However, the final paragraph of Section III-A states that the estimated shadow residual r_hat_1 is directly added to z_s, which would yield z_s + r_hat_1 = 2 z_s - z_hat_1^0. This is not a shadow-free latent unless z_s equals z_0. Since no code is provided, a reader implementing the equations literally and a reader implementing the prose description will obtain different outputs, and only one of them could have produced the reported PSNR/LPIPS/FID numbers. Please state the exact final reconstruction unambiguously and release code or pseudocode for the sampling loop.
  2. [Section III-A, Eqs. (6), (8), (9)] The noise schedule notation is internally inconsistent. Equation (6) writes z_t = sqrt(alpha_t) z'_t + sqrt(1 - alpha_t) epsilon, while Eq. (8) divides by sqrt(alpha_bar_t) and Eq. (9) rearranges the terms as if z_t were sqrt(alpha_bar_t) z'_t + sqrt(1 - alpha_bar_t) epsilon. In standard DDPM notation, alpha_bar_t is the cumulative product of the alpha coefficients, not the sum described in the text. As written, Eq. (9) does not follow from Eq. (6). Please define alpha_t and alpha_bar_t consistently and correct Eq. (6) and Eq. (13) so that the derivation is valid.
  3. [Section IV, Tables I-III] The paper reports only single-run numbers with no error bars or statistical significance tests, and it omits key sampling details such as the number of inference steps, any classifier-free guidance setting, and random seeds. Since the central claim is that the method improves over state-of-the-art methods, please provide uncertainty estimates for at least the main comparisons and state all inference hyperparameters. The absence of released code makes these details particularly important for reproducibility.
  4. [Section III-A, Eq. (9)] The derivation of Eq. (9) assumes that the pretrained noise estimator remains accurate when its input latent is a mixture of Gaussian noise and a shadow residual. This transferability assumption is plausible but is not directly validated anywhere in the paper. The authors should provide a per-timestep analysis, for example by measuring the error between the predicted noise/residual components and their ground-truth values during the backward process, or by comparing results when the noise estimator is fine-tuned versus kept frozen. Such an experiment would directly test the load-bearing assumption that the residual can be separated from the noise using the pretrained backbone.
minor comments (6)
  1. [Section III-A, Eq. (8)] The sentence preceding Eq. (8) says that z_hat_t^0 represents the estimated noise at step t, but z_hat_t^0 is the estimated clean latent, not the estimated noise. This typo should be corrected.
  2. [Section III-B] The description of the self-enhancement strategy says 'When the probability threshold p is less than P' without defining p. Presumably p is a random variable sampled uniformly in [0,1]; please state this explicitly.
  3. [Table IV vs Table I] The overall SSIM for the full pipeline is reported as 0.965 in Table I and 0.964 in Table IV. Please reconcile these numbers.
  4. [Section IV-D] The paper states that some methods were rerun with author-provided code and others were taken from benchmark [51], but it does not say which methods fall into which category. Please specify this so that the comparison is transparent.
  5. [Section IV-C] For the SRD dataset, the paper uses publicly available shadow masks from [30] for evaluation. It should state explicitly whether the same masks were used for all compared methods and for the reported PSNR-S/SSIM-S values.
  6. [Section IV-F] The label 'Ours-w/o EMA' appears in Table IV while the text uses 'Ours w/o EMA', and the header row contains 'Ours full pipline'. These typos should be fixed.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: residual-generation equations are algebraic identities and results are evaluated on independent benchmark splits.

full rationale

The paper's derivation chain is self-contained rather than circular. The modified forward process in Eqs. 5-6 defines the noisy latent as an interpolation between the shadow latent z_s and the shadow-free latent z_0 plus Gaussian noise. Eq. 8 solves for the network noise estimate, and Eq. 9 is a direct algebraic substitution showing that the estimated noise equals the true noise plus a residual-dependent term under the assumption that the network correctly predicts z_0. Eq. 12 is likewise an algebraic decomposition, not a fitted parameter disguised as a prediction. No constant is tuned to the reported PSNR, SSIM, LPIPS, or FID values, and the model is trained and evaluated on standard independent splits of ISTD+ and SRD against external SOTA methods. The cross-timestep self-enhancement strategy is an explicitly described self-training/data-augmentation mechanism, not a hidden reuse of test labels or target metrics. Citations such as RDDM [16] and ControlNet [9] are used as external building blocks, with clear differences stated, and no load-bearing argument reduces to an unverified self-citation. One non-circular concern is flagged: at the end of Section III-A, Eq. 10 defines r_hat_t = z_s - z_hat_t^0 and the prose says to add r_hat_1 to z_s, giving 2z_s - z_hat_1^0, whereas Eq. 11 with beta_bar_0 = 0 implies the final shadow-free latent should be z_s - r_hat_1 = z_hat_1^0. This is an internal sign inconsistency and a reproducibility issue, but it is not a circularity because it does not make the claimed result equivalent to its input by construction. Overall, the derivation is not circular, so the score is 0.

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

The method introduces a residual schedule and training hyperparameters, but no new physical entities or forces. The central claim rests on the transferability of a pre-trained diffusion backbone to a shifted input distribution and on standard DDIM algebra, with the notation in the paper being the main source of confusion.

free parameters (3)
  • shadow residual schedule β = linear from 0 to 1 over T
    The forward diffusion interpolates between shadow and shadow-free latents; the schedule shape is chosen ad hoc following RDDM, and the paper does not justify it beyond empirical performance.
  • EMA smoothing factor η = 0.999
    Used in cross-timestep self-enhancement to update the copy network; set empirically.
  • self-enhancement probability P = 0.2
    Probability of using network-generated pseudo-inputs during training; chosen empirically.
assumptions (3)
  • domain assumption The pre-trained Paint-By-Example inpainting model provides a valid generative latent prior for shadow-free image content.
    The method relies on the pre-trained backbone's noise estimator to remain valid when input latents are mixtures of noise and shadow residuals; this is assumed without theoretical justification.
  • standard math The DDIM noise schedule relationships (Eq. 1-3) hold for the modified forward process.
    The derivation of Eq. 9 assumes ᾱ_t is the cumulative noise coefficient and that Eq. 6 and Eq. 8 are consistent; the paper's notation confuses α_t and ᾱ_t.
  • domain assumption The control network can be initialized from the pre-trained model and fine-tuned without degrading the frozen backbone's capabilities.
    The method adds a ControlNet-style branch and assumes it can learn residual estimation while the backbone remains fixed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Controlling the Latent Diffusion Model for Generative Image Shadow Removal via Residual Generation." pith.science (2026). https://pith.science/paper/Y4NRPVXO

@misc{pith2026241202322,
  author       = {Pith},
  title        = {Pith review of: Controlling the Latent Diffusion Model for Generative Image Shadow Removal via Residual Generation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/Y4NRPVXO}},
  note         = {Machine review of arXiv:2412.02322}
}
read the original abstract

Large-scale generative models have achieved remarkable advancements in various visual tasks, yet their application to shadow removal in images remains challenging. These models often generate diverse, realistic details without adequate focus on fidelity, failing to meet the crucial requirements of shadow removal, which necessitates precise preservation of image content. In contrast to prior approaches that aimed to regenerate shadow-free images from scratch, this paper utilizes diffusion models to generate and refine image residuals. This strategy fully uses the inherent detailed information within shadowed images, resulting in a more efficient and faithful reconstruction of shadow-free content. Additionally, to revent the accumulation of errors during the generation process, a crosstimestep self-enhancement training strategy is proposed. This strategy leverages the network itself to augment the training data, not only increasing the volume of data but also enabling the network to dynamically correct its generation trajectory, ensuring a more accurate and robust output. In addition, to address the loss of original details in the process of image encoding and decoding of large generative models, a content-preserved encoder-decoder structure is designed with a control mechanism and multi-scale skip connections to achieve high-fidelity shadow-free image reconstruction. Experimental results demonstrate that the proposed method can reproduce high-quality results based on a large latent diffusion prior and faithfully preserve the original contents in shadow regions.

Figures

Figures reproduced from arXiv: 2412.02322 by the authors.

Figure 1
Figure 1. Current SOTA algorithms still cannot completely remove complex [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Diffusion backward processes of different methods. (a) Denoising [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Flowchart of the training phase of the proposed method. [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: Flowchart of the inference (sampling) phase of the proposed method. The latent [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: The schematic illustration of our training strategy. [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: The structure of the proposed detail-preserving decoder. [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: Visual comparison of different methods without input masks on the ISTD+ dataset, with enlarged views of shadow edges for clearer contrast. [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 8
Figure 8. Figure 8: Visual comparison of different methods without input masks on the SRD dataset, with enlarged views of shadow edges for clearer contrast. [PITH_FULL_IMAGE:figures/full_fig_p009_8.png]
Figure 9
Figure 9. Figure 9: Visual comparison of various methods with input masks on the ISTD+ dataset. [PITH_FULL_IMAGE:figures/full_fig_p010_9.png]
Figure 10
Figure 10. Figure 10: Comparison of visual results with and without the proposed detail [PITH_FULL_IMAGE:figures/full_fig_p011_10.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Prompt-Aware Controllable Shadow Removal

    cs.CV 2025-01 conditional novelty 6.0 of 10

    A prompt-aware network, PACSRNet, removes the shadow of a user-specified subject using a dot, line, or subject mask, and the authors introduce a new dataset for this task.

Reference graph

Works this paper leans on

53 extracted references · 25 canonical work pages · cited by 1 Pith paper

  1. [1]

    Shadowformer: Global context helps shadow removal,

    L. Guo, S. Huang, D. Liu, H. Cheng, and B. Wen, “Shadowformer: Global context helps shadow removal,” in Proceedings of the AAAI Conference on Artificial Intelligence , vol. 37, no. 1, 2023, pp. 710–718. JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 12

  2. [2]

    Leveraging inpainting for single-image shadow removal,

    X. Li, Q. Guo, R. Abdelfattah, D. Lin, W. Feng, I. Tsang, and S. Wang, “Leveraging inpainting for single-image shadow removal,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2023, pp. 13 055–13 064

  3. [3]

    A shadow imaging bilinear model and three-branch residual network for shadow removal,

    J. Liu, Q. Wang, H. Fan, J. Tian, and Y . Tang, “A shadow imaging bilinear model and three-branch residual network for shadow removal,” IEEE Transactions on Neural Networks and Learning Systems, pp. 1–15, 2023

  4. [4]

    Homoformer: Homogenized transformer for image shadow removal,

    J. Xiao, X. Fu, Y . Zhu, D. Li, J. Huang, K. Zhu, and Z.-J. Zha, “Homoformer: Homogenized transformer for image shadow removal,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2024, pp. 25 617–25 626

  5. [5]

    Shadow removal via shadow image decompo- sition,

    H. Le and D. Samaras, “Shadow removal via shadow image decompo- sition,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2019, pp. 8578–8587

  6. [6]

    Denoising diffusion probabilistic models,

    J. Ho, A. Jain, and P. Abbeel, “Denoising diffusion probabilistic models,” Advances in neural information processing systems , vol. 33, pp. 6840– 6851, 2020

  7. [7]

    Denoising diffusion implicit models,

    J. Song, C. Meng, and S. Ermon, “Denoising diffusion implicit models,” arXiv preprint arXiv:2010.02502 , 2020

  8. [8]

    High- resolution image synthesis with latent diffusion models,

    R. Rombach, A. Blattmann, D. Lorenz, P. Esser, and B. Ommer, “High- resolution image synthesis with latent diffusion models,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2022, pp. 10 684–10 695

Show all 53 references
  1. [9]

    Adding conditional control to text-to-image diffusion models,

    L. Zhang, A. Rao, and M. Agrawala, “Adding conditional control to text-to-image diffusion models,” in Proceedings of the IEEE/CVF International Conference on Computer Vision , 2023, pp. 3836–3847

  2. [10]

    Sgdm: An adaptive style- guided diffusion model for personalized text to image generation,

    Y . Xu, X. Xu, H. Gao, and F. Xiao, “Sgdm: An adaptive style- guided diffusion model for personalized text to image generation,” IEEE Transactions on Multimedia , vol. 26, pp. 9804–9813, 2024

  3. [11]

    Animediff: Customized image generation of anime characters using diffusion model,

    Y . Jiang, Q. Liu, D. Chen, L. Yuan, and Y . Fu, “Animediff: Customized image generation of anime characters using diffusion model,” IEEE Transactions on Multimedia , pp. 1–13, 2024

  4. [12]

    Diffbir: Towards blind image restoration with generative diffusion prior,

    X. Lin, J. He, Z. Chen, Z. Lyu, B. Dai, F. Yu, W. Ouyang, Y . Qiao, and C. Dong, “Diffbir: Towards blind image restoration with generative diffusion prior,” arXiv preprint arXiv:2308.15070 , 2023

  5. [13]

    Diffusion model for generative image denoising,

    Y . Xie, M. Yuan, B. Dong, and Q. Li, “Diffusion model for generative image denoising,” arXiv preprint arXiv:2302.02398 , 2023

  6. [14]

    Denoising diffusion restoration models,

    B. Kawar, M. Elad, S. Ermon, and J. Song, “Denoising diffusion restoration models,” Advances in Neural Information Processing Sys- tems, vol. 35, pp. 23 593–23 606, 2022

  7. [15]

    Controlling vision-language models for universal image restoration,

    Z. Luo, F. K. Gustafsson, Z. Zhao, J. Sj ¨olund, and T. B. Sch ¨on, “Controlling vision-language models for universal image restoration,” arXiv preprint arXiv:2310.01018 , vol. 3, no. 8, 2023

  8. [16]

    Residual de- noising diffusion models,

    J. Liu, Q. Wang, H. Fan, Y . Wang, Y . Tang, and L. Qu, “Residual de- noising diffusion models,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2024, pp. 2773–2783

  9. [17]

    The open images dataset v4: Unified image classification, object detection, and visual relationship detection at scale,

    A. Kuznetsova, H. Rom, N. Alldrin, J. Uijlings, I. Krasin, J. Pont-Tuset, S. Kamali, S. Popov, M. Malloci, A. Kolesnikovet al., “The open images dataset v4: Unified image classification, object detection, and visual relationship detection at scale,” International journal of co...

  10. [18]

    Laion- 5b: An open large-scale dataset for training next generation image-text models,

    C. Schuhmann, R. Beaumont, R. Vencu, C. Gordon, R. Wightman, M. Cherti, T. Coombes, A. Katta, C. Mullis, M. Wortsmanet al., “Laion- 5b: An open large-scale dataset for training next generation image-text models,” Advances in Neural Information Processing Systems , vol. 35, pp....

  11. [19]

    Shadow remover: Image shadow removal based on illumination recovering optimization,

    L. Zhang, Q. Zhang, and C. Xiao, “Shadow remover: Image shadow removal based on illumination recovering optimization,” IEEE Transac- tions on Image Processing , vol. 24, no. 11, pp. 4623–4636, 2015

  12. [20]

    The shadow meets the mask: Pyramid-based shadow removal,

    Y . Shor and D. Lischinski, “The shadow meets the mask: Pyramid-based shadow removal,” in Computer Graphics Forum, vol. 27, 2008, pp. 577– 586

  13. [21]

    On the removal of shadows from images,

    G. D. Finlayson, S. D. Hordley, C. Lu, and M. S. Drew, “On the removal of shadows from images,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 28, no. 1, pp. 59–68, 2005

  14. [22]

    Learning to remove soft shadows,

    M. Gryka, M. Terry, and G. J. Brostow, “Learning to remove soft shadows,” ACM Transactions on Graphics , vol. 34, no. 5, pp. 1–15, 2015

  15. [23]

    Paired regions for shadow detection and removal,

    R. Guo, Q. Dai, and D. Hoiem, “Paired regions for shadow detection and removal,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 35, no. 12, pp. 2956–2967, 2012

  16. [24]

    Leave-one-out kernel optimization for shadow detection and removal,

    T. F. Y . Vicente, M. Hoai, and D. Samaras, “Leave-one-out kernel optimization for shadow detection and removal,” IEEE Transactions on Pattern Analysis and Machine Intelligence , vol. 40, no. 3, pp. 682–695, 2017

  17. [25]

    Stacked conditional generative adversarial networks for jointly learning shadow detection and shadow removal,

    J. Wang, X. Li, and J. Yang, “Stacked conditional generative adversarial networks for jointly learning shadow detection and shadow removal,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 1788–1797

  18. [26]

    Auto-exposure fusion for single-image shadow removal,

    L. Fu, C. Zhou, Q. Guo, F. Juefei-Xu, H. Yu, W. Feng, Y . Liu, and S. Wang, “Auto-exposure fusion for single-image shadow removal,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2021, pp. 10 571–10 580

  19. [27]

    A boundary-aware network for shadow removal,

    K. Niu, Y . Liu, E. Wu, and G. Xing, “A boundary-aware network for shadow removal,” IEEE Transactions on Multimedia, vol. 25, pp. 6782– 6793, 2023

  20. [28]

    Des3: Adaptive attention-driven self and soft shadow removal using vit similarity,

    Y . Jin, W. Ye, W. Yang, Y . Yuan, and R. T. Tan, “Des3: Adaptive attention-driven self and soft shadow removal using vit similarity,” in Proceedings of the AAAI Conference on Artificial Intelligence , vol. 38, no. 3, 2024, pp. 2634–2642

  21. [29]

    Mask-shadowgan: Learn- ing to remove shadows from unpaired data,

    X. Hu, Y . Jiang, C.-W. Fu, and P.-A. Heng, “Mask-shadowgan: Learn- ing to remove shadows from unpaired data,” in Proceedings of the IEEE/CVF international conference on computer vision, 2019, pp. 2472– 2481

  22. [30]

    Towards ghost-free shadow removal via dual hierarchical aggregation network and shadow matting gan,

    X. Cun, C.-M. Pun, and C. Shi, “Towards ghost-free shadow removal via dual hierarchical aggregation network and shadow matting gan,” in Proceedings of the AAAI Conference on Artificial Intelligence , vol. 34, no. 07, 2020, pp. 10 680–10 687

  23. [31]

    Dc-shadownet: Single-image hard and soft shadow removal using unsupervised domain-classifier guided network,

    Y . Jin, A. Sharma, and R. T. Tan, “Dc-shadownet: Single-image hard and soft shadow removal using unsupervised domain-classifier guided network,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021, pp. 5027–5036

  24. [32]

    A decoupled multi- task network for shadow removal,

    J. Liu, Q. Wang, H. Fan, W. Li, L. Qu, and Y . Tang, “A decoupled multi- task network for shadow removal,” IEEE Transactions on Multimedia , vol. 25, pp. 9449–9463, 2023

  25. [33]

    Mmginpainting: Multi-modality guided image inpainting based on diffusion models,

    C. Zhang, W. Yang, X. Li, and H. Han, “Mmginpainting: Multi-modality guided image inpainting based on diffusion models,” IEEE Transactions on Multimedia, vol. 26, pp. 8811–8823, 2024

  26. [34]

    Wavedm: Wavelet-based diffusion models for image restoration,

    Y . Huang, J. Huang, J. Liu, M. Yan, Y . Dong, J. Lv, C. Chen, and S. Chen, “Wavedm: Wavelet-based diffusion models for image restoration,” IEEE Transactions on Multimedia, vol. 26, pp. 7058–7073, 2024

  27. [35]

    Shadowdiffusion: When degradation prior meets diffusion model for shadow removal,

    L. Guo, C. Wang, W. Yang, S. Huang, Y . Wang, H. Pfister, and B. Wen, “Shadowdiffusion: When degradation prior meets diffusion model for shadow removal,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2023, pp. 14 049–14 058

  28. [36]

    Latent feature-guided diffusion models for shadow removal,

    K. Mei, L. Figueroa, Z. Lin, Z. Ding, S. Cohen, and V . M. Patel, “Latent feature-guided diffusion models for shadow removal,” in Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision , 2024, pp. 4313–4322

  29. [37]

    Scaling up to excellence: Practicing model scaling for photo- realistic image restoration in the wild,

    F. Yu, J. Gu, Z. Li, J. Hu, X. Kong, X. Wang, J. He, Y . Qiao, and C. Dong, “Scaling up to excellence: Practicing model scaling for photo- realistic image restoration in the wild,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2024, pp....

  30. [38]

    Seesr: Towards semantics-aware real-world image super-resolution,

    R. Wu, T. Yang, L. Sun, Z. Zhang, S. Li, and L. Zhang, “Seesr: Towards semantics-aware real-world image super-resolution,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2024, pp. 25 456–25 467

  31. [39]

    Paint by example: Exemplar-based image editing with diffusion models,

    B. Yang, S. Gu, B. Zhang, T. Zhang, X. Chen, X. Sun, D. Chen, and F. Wen, “Paint by example: Exemplar-based image editing with diffusion models,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2023, pp. 18 381–18 391

  32. [40]

    Momentum contrast for unsupervised visual representation learning,

    K. He, H. Fan, Y . Wu, S. Xie, and R. Girshick, “Momentum contrast for unsupervised visual representation learning,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2020, pp. 9729–9738

  33. [41]

    Taming transformers for high- resolution image synthesis,

    P. Esser, R. Rombach, and B. Ommer, “Taming transformers for high- resolution image synthesis,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2021, pp. 12 873–12 883

  34. [42]

    Designing a better asymmetric vqgan for stablediffusion,

    Z. Zhu, X. Feng, D. Chen, J. Bao, L. Wang, Y . Chen, L. Yuan, and G. Hua, “Designing a better asymmetric vqgan for stablediffusion,” arXiv preprint arXiv:2306.04632 , 2023

  35. [43]

    Deformable convnets v2: More deformable, better results,

    X. Zhu, H. Hu, S. Lin, and J. Dai, “Deformable convnets v2: More deformable, better results,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2019, pp. 9308–9316

  36. [44]

    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. Clark et al., “Learning transferable visual models from natural language supervision,” in International conference on machine learning . PMLR, 2021, pp. 8748–8763. JOURNAL OF...

  37. [45]

    Physics-based shadow image decomposition for shadow removal,

    H. Le and D. Samaras, “Physics-based shadow image decomposition for shadow removal,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 44, no. 12, pp. 9088–9101, 2021

  38. [46]

    Deshadownet: A multi- context embedding deep network for shadow removal,

    L. Qu, J. Tian, S. He, Y . Tang, and R. W. Lau, “Deshadownet: A multi- context embedding deep network for shadow removal,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2017, pp. 4067–4075

  39. [47]

    The unreasonable effectiveness of deep features as a perceptual metric,

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

  40. [48]

    Gans trained by a two time-scale update rule converge to a local nash equilibrium,

    M. Heusel, H. Ramsauer, T. Unterthiner, B. Nessler, and S. Hochreiter, “Gans trained by a two time-scale update rule converge to a local nash equilibrium,” Advances in neural information processing systems , vol. 30, 2017

  41. [49]

    Direction-aware spatial context features for shadow detection,

    X. Hu, L. Zhu, C.-W. Fu, J. Qin, and P.-A. Heng, “Direction-aware spatial context features for shadow detection,” in Proceedings of the IEEE conference on computer vision and pattern recognition , 2018, pp. 7454–7462

  42. [50]

    Shadow removal by a lightness-guided network with training on unpaired data,

    Z. Liu, H. Yin, Y . Mi, M. Pu, and S. Wang, “Shadow removal by a lightness-guided network with training on unpaired data,” IEEE Transactions on Image Processing , vol. 30, pp. 1853–1865, 2021

  43. [51]

    Unveiling deep shadows: A survey on image and video shadow detection, removal, and generation in the era of deep learning,

    X. Hu, Z. Xing, T. Wang, C.-W. Fu, and P.-A. Heng, “Unveiling deep shadows: A survey on image and video shadow detection, removal, and generation in the era of deep learning,”arXiv preprint arXiv:2409.02108, 2024

  44. [52]

    Style- guided shadow removal,

    J. Wan, H. Yin, Z. Wu, X. Wu, Y . Liu, and S. Wang, “Style- guided shadow removal,” in European Conference on Computer Vision . Springer, 2022, pp. 361–378

  45. [53]

    Bijective mapping network for shadow removal,

    Y . Zhu, J. Huang, X. Fu, F. Zhao, Q. Sun, and Z.-J. Zha, “Bijective mapping network for shadow removal,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2022, pp. 5627–5636

Pith tools

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