Pith. sign in

REVIEW 2 major objections 7 minor 86 references

NTIRE 2025 Image Shadow Removal Challenge Report

T0 review · 2 major / 7 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read Re-aligning the WSRD+ shadow dataset raised baseline PSNR by about 2 dB, and the top method won both fidelity and perceptual tracks.

desk verdict A standard NTIRE challenge report with one genuinely new but under-validated data-alignment step; the perceptual leaderboard is too fragile to be the load-bearing result. read the letter →

arxiv 2506.15524 v1 pith:CW5OQBUK submitted 2025-06-18 cs.CV

Florin-Alexandru Vasluianu , Tim Seizinger , Zhuyun Zhou , Cailian Chen , Zongwei Wu , Radu Timofte , Mingjia Li , Jin Hu
show 74 more authors
Hainuo Wang Hengxing Liu Jiarui Wang Qiming Hu Xiaojie Guo Xin Lu Jiarong Yang Yuanfei Bao Anya Hu Zihao Fan Kunyu Wang Jie Xiao Xi Wang Xueyang Fu Zheng-Jun Zha Yu-Fan Lin Chia-Ming Lee Chih-Chung Hsu Xingbo Wang Dong Li Yuxu Chen Bin Chen Yuanbo Zhou Yuanbin Chen Hongwei Wang Jiannan Lin Qinquan Gao Tong Tong Zhao Zhang Yanyan Wei Wei Dong Han Zhou Seyed Amirreza Mousavi Jun Chen Haobo Liang Jiajie Jing Junyu Li Yan Yang Seoyeon Lee Chaewon Kim Ziyu Feng Shidi Chen Bowen Luan Zewen Chen Vijayalaxmi Ashok Aralikatti G Gyaneshwar Rao Nikhil Akalwadi Chaitra Desai Ramesh Ashok Tabib Uma Mudenagudi Anas M. Ali Bilel Benjdira Wadii Boulila Alexandru Brateanu Cosmin Ancuti Tanmay Chaturvedi Manish Kumar Anmol Srivastav Daksh Trivedi Shashwat Thakur Kishor Upla Zeyu Xiao Zhuoyuan Li Boda Zhou Shashank Shekhar Kele Xu Qisheng Xu Zijian Gao Tianjiao Wan Suiyi Zhao Bo Wang Yan Luo Mingshen Wang Yilin Zhang
This is my paper
classification cs.CV
keywords shadowremovalimagerestorationbenchmarkdatasetalignmentuserstudymeanopinionscoreevaluationmetrichigh-resolutionimages
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 report presents a shadow removal challenge run on WSRD+, a benchmark of high-resolution photographs containing both self-shadows and cast shadows from multiple directional lights. Its central claim is that a new two-stage alignment procedure—keypoint-based homography followed by contour-driven affinity refinement—substantially improves the quality of the benchmark's ground-truth pairs, raising measured PSNR by about 2 dB on the baseline data. With this better-aligned data, the report says, 17 submitted methods could be evaluated fairly, and the same method, X-Shadow, ranked first in both the fidelity track (PSNR, SSIM, LPIPS) and the perceptual track (a user-study mean opinion score).

What carries the argument

The central object is WSRD+, a high-resolution shadow-removal dataset whose capture setup uses multiple directional lights of varying sizes, positions, and distances to create umbra and penumbra variation in self- and cast shadows. The load-bearing mechanism is the double-stage alignment pipeline introduced in this report: a first pass estimates a homography from keypoint matches, and a second pass refines it with a gradient-descent affinity estimate based on object contour geometry, specifically targeting mismatches in severely underexposed image regions. The report credits this pipeline for the ~2 dB improvement and for letting participants omit their own alignment compensation.

What would settle it

Run the same user study on all 75 test images (or on a random subset stratified by shadow complexity) and check whether the MOS ordering of the 17 submissions is preserved; if the ordering changes materially, the perceptual track's ranking is an artifact of the ten selected images.

Watch

Extended reading notes

Core claim

The paper's central discovery, stated on its own terms, is that the WSRD+ re-alignment protocol converts a benchmark whose pixel mismatch had been a major obstacle into one where modern deep shadow-removal systems show a clear and quantified performance gain. The organizers report an approximately 2 dB PSNR increase for the baseline data after the two-stage alignment, and attribute the improved submissions to this data-quality fix. On the final 75-image test split, the 17 valid submissions span PSNR from roughly 14.9 to 25.9, SSIM from 0.55 to 0.84, LPIPS from 0.08 to 0.66, and MOS from 0.8 to 8.75, with the top four teams separated by less than 0.6 PSNR and 0.75 MOS.

Load-bearing premise

The perceptual track's ranking depends on a user study of ten test images that the organizers selected because they are challenging, and the report provides no evidence that those ten represent the full 75-image test set.

Editorial extensions

If this is right

  • Teams could skip custom misalignment compensation and concentrate on shadow-removal architecture, which is why the report frames the ~2 dB lift as opening the path to higher-quality results.
  • The leading solution, X-Shadow, shows that feeding intrinsic hints (albedo, irradiance, normal, depth) into a two-stage low-then-high-resolution network is a strong recipe for mask-free shadow removal.
  • The top five systems are close in both fidelity (PSNR 24.8–25.9) and perception (MOS 7.98–8.75), so the leaderboard separates the field but not dramatically.
  • The reported correlation between the PSNR/SSIM/LPIPS ordering and the user-study ordering suggests that on this benchmark, fidelity metrics and human perception point the same way.

Reading between the lines

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

  • The MOS ranking uses only ten hand-picked 'challenging' test images; extending the user study to the full 75-image split could change the perceptual ordering, since the report gives no rater-agreement or significance statistics.
  • The ~2 dB gain is measured against the previous alignment protocol; an independent alignment-error metric (e.g., optical-flow consistency between input and reference) would show how much of the improvement is geometric correction versus an artifact of the specific matching method.
  • Because every team trained on the same aligned pairs, the challenge does not probe robustness to naturally misaligned or in-the-wild shadow photos; a test set with deliberately unaligned pairs would reveal which methods depend on the alignment fix.
  • The winner's reliance on auxiliary intrinsic predictors suggests shadow removal could be reframed as a joint scene-understanding task; ablating each hint (albedo, irradiance, normal, depth) would test that dependency directly.
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

2 major / 7 minor

Summary. The paper reports the organization and results of the NTIRE 2025 Image Shadow Removal Challenge. Two tracks are defined: a fidelity track based on PSNR, SSIM, LPIPS and a parameter-efficiency measure, and a perceptual track based on mean opinion scores from a user study. Seventeen teams are evaluated on 75 test images from the WSRD+ dataset. The paper presents Table 1 with per-team metrics and ranks, a figure with sample outputs, short descriptions of each submitted method, and a claim that a revised two-stage alignment of WSRD+ improves baseline PSNR by approximately 2 dB.

Significance. The paper provides a current snapshot of the state of the art in shadow removal under a common benchmark, which is useful to the community as a reference point for method comparison. It also documents a potentially valuable improvement to the WSRD+ dataset alignment. The evaluation is anchored to an externally defined test split with private ground truth, which limits circularity, and the paper links to a public dataset repository. However, the value of the paper as a benchmark report depends on the reliability and transparency of the two leaderboards; the perceptual leaderboard currently lacks statistical grounding, and the fidelity aggregation rule is incompletely specified.

major comments (2)
  1. [Section 5 / Table 1] The perceptual-track ranking rests entirely on MOS values computed from a user study on only 10 organizer-selected test images (indices 3, 4, 15, 17, 27, 39, 41, 55, 70, 73). The paper reports no number of raters, no per-image MOS values, no inter-rater agreement statistic, and no confidence intervals or significance tests. Adjacent top ranks differ by as little as 0.02 (FusionShadowRemoval 8.00 vs. GLHF 7.98) and by 0.10 (ACVLab 8.10 vs. LUMOS 8.20), while the stated grading scale has 0.5 increments; with a small number of images and raters, a single changed rating can flip such differences. Because the perceptual track is one of the two official rankings and the conclusion explicitly invokes the "rankings of the conducted user study," the paper should either add statistical support (rater count, per-image scores, bootstrap confidence intervals, or a significance test between adjacent ranks) or present the MOS ordering as preliminary rather than as a definitive perceptual leaderboard.
  2. [Section 3 / Table 1] The fidelity-track ranking rule is not fully specified. The text says PSNR, SSIM, and LPIPS are weighted equally and that "for solutions of similar fidelity, lower complexity approaches are favored through a parameter efficiency metric," but it does not define the efficiency metric, the normalization of the three image-quality metrics, or the tie-breaking procedure. Without this information, the reported Fidelity Ranks in Table 1 (for example, X-Shadow ranked first despite a lower PSNR than ACVLab and LUMOS and having 235 M parameters) cannot be independently reproduced or audited. Please provide the exact aggregation formula, including any normalization and the parameter-efficiency term, or cite a public specification that contains it.
minor comments (7)
  1. [Section 2] The claim of an "approximately 2 dB" PSNR increase from the double-stage alignment is not accompanied by any experiment, table, or baseline definition; please either provide the supporting comparison or remove the quantitative claim.
  2. [Section 7.1] DepthAnythingV2 is cited as [53], but reference [53] is the SAM 2 paper; a separate reference for DepthAnythingV2 appears to be missing.
  3. [Section 4] The text reads "a set of 75 input images input images"; the duplicate phrase should be corrected.
  4. [Section 7.11] The text reads "TIn the refinement module"; this should be "In the refinement module."
  5. [Section 7.14] The text reads "Team CV SVNIT a introduces"; the stray "a" should be removed.
  6. [Table 1] For GroupNo9, the Params. entry is a dash; please use "n/a" or state explicitly that the parameter count was not reported.
  7. [Section 3] The metric is referred to as "Structured Similarity Index (SSIM)"; the standard name is "Structural Similarity Index Measure."

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the report is a challenge benchmark whose rankings are anchored to external WSRD+ ground truth; self-citations are provenance, and the 2 dB alignment claim and small MOS study are support/robustness concerns, not circular reductions.

full rationale

This is a challenge report, not a derivation of a predicted quantity from first principles. The load-bearing content is the ranking of 17 submitted methods on the WSRD+ test split, evaluated against private ground truth using PSNR, SSIM, LPIPS, and a user-study MOS. None of these evaluations are defined in terms of the methods' outputs in a way that forces the rankings by construction. The organizers' self-citations to WSRD [62], prior NTIRE shadow removal reports [63, 65], and the ambient lighting normalization extension [64] are used as dataset provenance and architectural context, not as unverified premises that uniquely force the winner or the reported numbers. The Section 2 statement that the double-stage alignment 'resulted in a notable increase in PSNR of approximately 2 dB for the baseline data' is an unsupported quantitative claim without a shown measurement protocol, but it is not circular: it is an empirical preprocessing claim, not a fitted parameter later renamed a prediction. The perceptual track uses MOS from only 10 organizer-selected images (indices 3, 4, 15, 17, 27, 39, 41, 55, 70, 73) with no rater counts, agreement metrics, or confidence intervals, and adjacent top MOS values differ by as little as 0.02; this is a statistical-validity weakness that could affect the reported ordering, but it does not make the ranking equivalent to its inputs by definition. The paper also honestly states its grading scale and the selection criterion ('chosen given the complexity of the shadow scenario'), so the limitation is disclosed in-scope. No self-definitional step, fitted-input-called-prediction step, load-bearing self-citation chain, uniqueness import, ansatz smuggling, or renaming of a known result was found. The central claims remain anchored to external ground truth and independent submissions, so the appropriate finding is no significant circularity.

Assumptions & free parameters 1 free parameters · 3 assumptions · 0 invented entities

The report introduces no theoretical entities. Its load-bearing assumptions are the validity of standard image quality metrics, the correctness of the aligned ground truth, and the representativeness of the 10-image user study.

free parameters (1)
  • User study subset = 10 images, indices 3,4,15,17,27,39,41,55,70,73
    Hand-picked by the organizers as challenging samples; the perceptual ranking depends on this selection.
assumptions (3)
  • domain assumption PSNR, SSIM, and LPIPS are valid measures of reconstruction fidelity for shadow removal.
    The fidelity track ranking is defined as the unweighted average of these three metrics (Section 3).
  • domain assumption The WSRD+ ground-truth images are correctly aligned and free of illumination inconsistencies.
    The paper's alignment strategy is applied exactly to establish this, and the 2 dB claim assumes the aligned reference is the correct target.
  • domain assumption Mean opinion scores from expert raters on a 1-10 scale measure perceptual quality faithfully.
    The perceptual track uses MOS as the principal ranking criterion (Section 5).

how reviews work

0 comments
Cite this review

Pith. "Pith review of NTIRE 2025 Image Shadow Removal Challenge Report." pith.science (2026). https://pith.science/paper/CW5OQBUK

@misc{pith2026250615524,
  author       = {Pith},
  title        = {Pith review of: NTIRE 2025 Image Shadow Removal Challenge Report},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CW5OQBUK}},
  note         = {Machine review of arXiv:2506.15524}
}
read the original abstract

This work examines the findings of the NTIRE 2025 Shadow Removal Challenge. A total of 306 participants have registered, with 17 teams successfully submitting their solutions during the final evaluation phase. Following the last two editions, this challenge had two evaluation tracks: one focusing on reconstruction fidelity and the other on visual perception through a user study. Both tracks were evaluated with images from the WSRD+ dataset, simulating interactions between self- and cast-shadows with a large number of diverse objects, textures, and materials.

Figures

Figures reproduced from arXiv: 2506.15524 by the authors.

Figure 1
Figure 1. Equivalent samples from the NTIRE 2025 Image Shadow Removal Challenge test split, for Team X-Shadow, the winner of both [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. The overall architecture proposed by X-Shadow. In [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Overview of the Team LUMOS proposed EvenFormer. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (4 more)
Figure 7
Figure 7. Figure 7: The two stage L-Adapter architecture proposed by MI [PITH_FULL_IMAGE:figures/full_fig_p006_7.png]
Figure 8
Figure 8. Figure 8: A graphical representation of the solution proposed by [PITH_FULL_IMAGE:figures/full_fig_p006_8.png]
Figure 9
Figure 9. Figure 9: Overview of the OptiMalDiff architecture combining [PITH_FULL_IMAGE:figures/full_fig_p007_9.png]
Figure 10
Figure 10. Figure 10: Overview of the dual branch architecture proposed by [PITH_FULL_IMAGE:figures/full_fig_p007_10.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

86 extracted references · 67 canonical work pages

  1. [1]

    Kolmogorov-arnold networks in transformer attention for low-light image en- hancement

    Alexandru Brateanu and Raul Balmez. Kolmogorov-arnold networks in transformer attention for low-light image en- hancement. In2024 International Symposium on Electronics and Telecommunications (ISETC), pages 1–4, 2024. 8

    2024

  2. [2]

    LYT-NET: Lightweight YUV Transformer-based Network for Low-light Image Enhancement

    Alexandru Brateanu, Raul Balmez, Adrian Avram, Ciprian Orhei, and Ancuti Cosmin. Lyt-net: Lightweight yuv transformer-based network for low-light image enhance- ment.arXiv preprint arXiv:2401.15204, 2024. 8

    work page Pith review arXiv 2024

  3. [3]

    Enhancing low-light images with kolmogorov–arnold networks in transformer attention

    Alexandru Brateanu, Raul Balmez, Ciprian Orhei, Cosmin Ancuti, and Codruta Ancuti. Enhancing low-light images with kolmogorov–arnold networks in transformer attention. Sensors, 25(2), 2025. 8

    2025

  4. [4]

    Retinexformer: One-stage retinex- based transformer for low-light image enhancement

    Yuanhao Cai, Hao Bian, Jing Lin, Haoqian Wang, Radu Tim- ofte, and Yulun Zhang. Retinexformer: One-stage retinex- based transformer for low-light image enhancement. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 12504–12513, 2023. 7

    2023

  5. [5]

    Transunet: Rethinking the u-net archi- tecture design for medical image segmentation through the lens of transformers.Medical Image Analysis, page 103280,

    Jieneng Chen, Jieru Mei, Xianhang Li, Yongyi Lu, Qihang Yu, Qingyue Wei, Xiangde Luo, Yutong Xie, Ehsan Adeli, Yan Wang, et al. Transunet: Rethinking the u-net archi- tecture design for medical image segmentation through the lens of transformers.Medical Image Analysis, page 103280,

  6. [6]

    Simple baselines for image restoration.arXiv preprint arXiv:2204.04676, 2022

    Liangyu Chen, Xiaojie Chu, Xiangyu Zhang, and Jian Sun. Simple baselines for image restoration.arXiv preprint arXiv:2204.04676, 2022. 5, 6, 8

    arXiv 2022

  7. [7]

    Frequency-aware feature fusion for dense image prediction.IEEE Transactions on Pattern Anal- ysis and Machine Intelligence, 46(12):10763–10780, 2024

    Linwei Chen, Ying Fu, Lin Gu, Chenggang Yan, Tatsuya Harada, and Gao Huang. Frequency-aware feature fusion for dense image prediction.IEEE Transactions on Pattern Anal- ysis and Machine Intelligence, 46(12):10763–10780, 2024. 5

    2024

  8. [8]

    NTIRE 2025 challenge on image super-resolution (×4): Methods and results

    Zheng Chen, Kai Liu, Jue Gong, Jingkai Wang, Lei Sun, Zongwei Wu, Radu Timofte, Yulun Zhang, et al. NTIRE 2025 challenge on image super-resolution (×4): Methods and results. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Work- shops, 2025. 2

    2025

Show all 86 references
  1. [9]

    NTIRE 2025 challenge on real-world face restoration: Methods and results

    Zheng Chen, Jingkai Wang, Kai Liu, Jue Gong, Lei Sun, Zongwei Wu, Radu Timofte, Yulun Zhang, et al. NTIRE 2025 challenge on real-world face restoration: Methods and results. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Work- shops, 2025. 2

    2025

  2. [10]

    NTIRE 2025 challenge on raw image restoration and super-resolution

    Marcos Conde, Radu Timofte, et al. NTIRE 2025 challenge on raw image restoration and super-resolution. InProceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2025. 2

    2025

  3. [11]

    Raw image reconstruc- tion from RGB on smartphones

    Marcos Conde, Radu Timofte, et al. Raw image reconstruc- tion from RGB on smartphones. NTIRE 2025 challenge re- port. InProceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition (CVPR) Workshops,

    2025

  4. [12]

    Diffusion schr¨odinger bridge with applications to score-based generative modeling.Advances in Neural In- formation Processing Systems, 34:17695–17709, 2021

    Valentin De Bortoli, James Thornton, Jeremy Heng, and Ar- naud Doucet. Diffusion schr¨odinger bridge with applications to score-based generative modeling.Advances in Neural In- formation Processing Systems, 34:17695–17709, 2021. 7

    2021

  5. [13]

    Imagenet: A large-scale hierarchical image database

    Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. Imagenet: A large-scale hierarchical image database. InCVPR, 2009. 6

    2009

  6. [14]

    Optimal transport losses and sinkhorn algorithm with general convex regular- ization.arXiv preprint arXiv:2007.00976, 2020

    Simone Di Marino and Augusto Gerolin. Optimal transport losses and sinkhorn algorithm with general convex regular- ization.arXiv preprint arXiv:2007.00976, 2020. 7

    2007 arXiv

  7. [15]

    Argan: Attentive recurrent generative adversarial network for shadow detection and removal

    Bin Ding, Chengjiang Long, Ling Zhang, and Chunxia Xiao. Argan: Attentive recurrent generative adversarial network for shadow detection and removal. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 10213–10222, 2019. 1

    2019

  8. [16]

    ShadowRefiner: Towards mask-free shadow removal via fast fourier transformer

    Wei Dong, Han Zhou, Yuqiong Tian, Jingke Sun, Xiaohong Liu, Guangtao Zhai, and Jun Chen. ShadowRefiner: Towards mask-free shadow removal via fast fourier transformer. In Proceedings of the IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition (CVPR) Workshops, 2024. 5, 6, 7

    2024

  9. [17]

    Wei Dong, Han Zhou, Yulun Zhang, Xiaohong Liu, and Jun Chen. Ecmamba: Consolidating selective state space model with retinex guidance for efficient multiple exposure correc- tion.Advances in Neural Information Processing Systems, 37:53438–53457, 2024. 7

    2024

  10. [18]

    NTIRE 2025 challenge on night photography rendering

    Egor Ershov, Sergey Korchagin, Alexei Khalin, Artyom Pan- shin, Arseniy Terekhin, Ekaterina Zaychenkova, Georgiy Lobarev, Vsevolod Plokhotnyuk, Denis Abramov, Elisey Zhdanov, Sofia Dorogova, Yasin Mamedov, Nikola Banic, Georgii Perevozchikov, Radu Timofte, et al. NTIRE 2025 ch...

    2025

  11. [19]

    Entropy minimization for shadow removal.International Journal of Computer Vision, 85(1):35–57, 2009

    Graham D Finlayson, Mark S Drew, and Cheng Lu. Entropy minimization for shadow removal.International Journal of Computer Vision, 85(1):35–57, 2009. 1

    2009

  12. [20]

    Removing shadows from images

    Graham D Finlayson, Steven D Hordley, and Mark S Drew. Removing shadows from images. InEuropean conference on computer vision, pages 823–836. Springer, 2002. 1

    2002

  13. [21]

    Dw-gan: A discrete wavelet transform gan for nonho- mogeneous dehazing

    Minghan Fu, Huan Liu, Yankun Yu, Jun Chen, and Keyan Wang. Dw-gan: A discrete wavelet transform gan for nonho- mogeneous dehazing. InProceedings of the IEEE/CVF con- ference on computer vision and pattern recognition, pages 203–212, 2021. 5

    2021

  14. [22]

    NTIRE 2025 challenge on cross-domain few-shot object detection: Methods and results

    Yuqian Fu, Xingyu Qiu, Bin Ren Yanwei Fu, Radu Timofte, Nicu Sebe, Ming-Hsuan Yang, Luc Van Gool, et al. NTIRE 2025 challenge on cross-domain few-shot object detection: Methods and results. InProceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition (...

    2025

  15. [23]

    Maciej Gryka, Michael Terry, and Gabriel J. Brostow. Learn- ing to remove soft shadows.ACM Transactions on Graphics,

  16. [24]

    Shadowformer: Global context helps image shadow removal.arXiv preprint arXiv:2302.01650, 2023

    Lanqing Guo, Siyu Huang, Ding Liu, Hao Cheng, and Bihan Wen. Shadowformer: Global context helps image shadow removal.arXiv preprint arXiv:2302.01650, 2023. 1

    2023 arXiv

  17. [25]

    Shadowd- iffusion: When degradation prior meets diffusion model for shadow removal

    Lanqing Guo, Chong Wang, Wenhan Yang, Siyu Huang, Yufei Wang, Hanspeter Pfister, and Bihan Wen. Shadowd- iffusion: When degradation prior meets diffusion model for shadow removal. InProceedings of the IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, pages 1404...

    2023

  18. [26]

    NTIRE 2025 challenge on text to image generation model quality assess- ment

    Shuhao Han, Haotian Fan, Fangyuan Kong, Wenjie Liao, Chunle Guo, Chongyi Li, Radu Timofte, et al. NTIRE 2025 challenge on text to image generation model quality assess- ment. InProceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition (CVPR) Workshops,

    2025

  19. [27]

    Masked autoencoders are scalable vision learners, 2021

    Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Doll´ar, and Ross Girshick. Masked autoencoders are scalable vision learners, 2021. 5

    2021

  20. [28]

    Denoising dif- fusion probabilistic models.Advances in neural information processing systems, 33:6840–6851, 2020

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising dif- fusion probabilistic models.Advances in neural information processing systems, 33:6840–6851, 2020. 1

    2020

  21. [29]

    Shadowhack: Hack- ing shadows via luminance-color divide and conquer.CoRR, abs/2412.02545, 2024

    Jin Hu, Mingjia Li, and Xiaojie Guo. Shadowhack: Hack- ing shadows via luminance-color divide and conquer.CoRR, abs/2412.02545, 2024. 3

    2024 arXiv

  22. [30]

    Mask-ShadowGAN: Learning to remove shadows from unpaired data

    Xiaowei Hu, Yitong Jiang, Chi-Wing Fu, and Pheng-Ann Heng. Mask-ShadowGAN: Learning to remove shadows from unpaired data. InICCV, 2019. 1

    2019

  23. [31]

    Image-to-image translation with conditional adver- sarial networks

    Phillip Isola, Jun-Yan Zhu, Tinghui Zhou, and Alexei A Efros. Image-to-image translation with conditional adver- sarial networks. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 1125–1134,

  24. [32]

    NTIRE 2025 challenge on video quality enhancement for video con- ferencing: Datasets, methods and results

    Varun Jain, Zongwei Wu, Quan Zou, Louis Florentin, Hen- rik Turbell, Sandeep Siddhartha, Radu Timofte, et al. NTIRE 2025 challenge on video quality enhancement for video con- ferencing: Datasets, methods and results. InProceedings of the IEEE/CVF Conference on Computer Vision ...

    2025

  25. [33]

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

    Yeying Jin, Aashish Sharma, and Robby 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, pages 5027–5036, 2021. 1, 2

    2021

  26. [34]

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

    Yeying Jin, Wei Ye, Wenhan Yang, Yuan Yuan, and Robby T Tan. Des3: Adaptive attention-driven self and soft shadow removal using vit similarity. InProceedings of the AAAI Conference on Artificial Intelligence, volume 38, pages 2634–2642, 2024. 1

    2024

  27. [35]

    Alex Krizhevsky, Ilya Sutskever, and Geoffrey E. Hinton. Imagenet classification with deep convolutional neural net- works.Commun. ACM, 60(6):84–90, may 2017. 3

    2017

  28. [36]

    The retinex theory of color vision.Scientific american, 237(6):108–129, 1977

    Edwin H Land. The retinex theory of color vision.Scientific american, 237(6):108–129, 1977. 7

    1977

  29. [37]

    Shadow removal via shadow image decomposition

    Hieu Le and Dimitris Samaras. Shadow removal via shadow image decomposition. InThe IEEE International Conference on Computer Vision (ICCV), October 2019. 1

    2019

  30. [38]

    From shadow segmentation to shadow removal

    Hieu Le and Dimitris Samaras. From shadow segmentation to shadow removal. InThe IEEE European Conference on Computer Vision (ECCV), August 2020. 1, 2

    2020

  31. [39]

    Physics-based shadow im- age decomposition for shadow removal.IEEE Transactions on Pattern Analysis and Machine Intelligence, 44(12):9088– 9101, 2021

    Hieu Le and Dimitris Samaras. Physics-based shadow im- age decomposition for shadow removal.IEEE Transactions on Pattern Analysis and Machine Intelligence, 44(12):9088– 9101, 2021. 4

    2021

  32. [40]

    NTIRE 2025 challenge on efficient burst hdr and restoration: Datasets, methods, and results

    Sangmin Lee, Eunpil Park, Angel Canelo, Hyunhee Park, Youngjo Kim, Hyungju Chun, Xin Jin, Chongyi Li, Chun-Le Guo, Radu Timofte, et al. NTIRE 2025 challenge on efficient burst hdr and restoration: Datasets, methods, and results. In Proceedings of the IEEE/CVF Conference on Com...

    2025

  33. [41]

    NTIRE 2025 challenge on day and night raindrop removal for dual-focused images: Methods and results

    Xin Li, Yeying Jin, Xin Jin, Zongwei Wu, Bingchen Li, Yufei Wang, Wenhan Yang, Yu Li, Zhibo Chen, Bihan Wen, Robby Tan, Radu Timofte, et al. NTIRE 2025 challenge on day and night raindrop removal for dual-focused images: Methods and results. InProceedings of the IEEE/CVF Confe...

    2025

  34. [42]

    NTIRE 2025 challenge on short-form ugc video quality assessment and enhancement: Kwaisr dataset and study

    Xin Li, Xijun Wang, Bingchen Li, Kun Yuan, Yizhen Shao, Suhang Yao, Ming Sun, Chao Zhou, Radu Timofte, and Zhibo Chen. NTIRE 2025 challenge on short-form ugc video quality assessment and enhancement: Kwaisr dataset and study. InProceedings of the IEEE/CVF Conference on Com- pu...

    2025

  35. [43]

    NTIRE 2025 challenge on short-form ugc video quality assessment and enhancement: Methods and results

    Xin Li, Kun Yuan, Bingchen Li, Fengbin Guan, Yizhen Shao, Zihao Yu, Xijun Wang, Yiting Lu, Wei Luo, Suhang Yao, Ming Sun, Chao Zhou, Zhibo Chen, Radu Timofte, et al. NTIRE 2025 challenge on short-form ugc video quality assessment and enhancement: Methods and results. InPro- ce...

    2025

  36. [44]

    NTIRE 2025 the 2nd restore any image model (RAIM) in the wild challenge

    Jie Liang, Radu Timofte, Qiaosi Yi, Zhengqiang Zhang, Shuaizheng Liu, Lingchen Sun, Rongyuan Wu, Xindong Zhang, Hui Zeng, Lei Zhang, et al. NTIRE 2025 the 2nd restore any image model (RAIM) in the wild challenge. In Proceedings of the IEEE/CVF Conference on Computer Vi- sion a...

    2025

  37. [45]

    NTIRE 2025 XGC quality assessment chal- lenge: Methods and results

    Xiaohong Liu, Xiongkuo Min, Qiang Hu, Xiaoyun Zhang, Jie Guo, et al. NTIRE 2025 XGC quality assessment chal- lenge: Methods and results. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2025. 2

    2025

  38. [46]

    NTIRE 2024 challenge on low light im- age enhancement: Methods and results

    Xiaoning Liu, Zongwei WU, Ao Li, Florin-Alexandru Vasluianu, Yulun Zhang, Shuhang Gu, Le Zhang, Ce Zhu, Radu Timofte, et al. NTIRE 2024 challenge on low light im- age enhancement: Methods and results. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Reco...

    2024

  39. [47]

    NTIRE 2025 challenge on low light image enhancement: Methods and results

    Xiaoning Liu, Zongwei Wu, Florin-Alexandru Vasluianu, Hailong Yan, Bin Ren, Yulun Zhang, Shuhang Gu, Le Zhang, Ce Zhu, Radu Timofte, et al. NTIRE 2025 challenge on low light image enhancement: Methods and results. In Proceedings of the IEEE/CVF Conference on Computer Vi- sion ...

    2025

  40. [48]

    Swin transformer: Hierarchical vision transformer using shifted windows

    Ze Liu, Yutong Lin, Yue Cao, Han Hu, Yixuan Wei, Zheng Zhang, Stephen Lin, and Baining Guo. Swin transformer: Hierarchical vision transformer using shifted windows. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2021. 7

    2021

  41. [49]

    A convnet for the 2020s

    Zhuang Liu, Hanzi Mao, Chao-Yuan Wu, Christoph Feicht- enhofer, Trevor Darrell, and Xie Saining. A convnet for the 2020s. InProceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition, 2022. 5, 6, 7, 8

    2022

  42. [50]

    Codalab competitions: An open source platform to organize scientific challenges.Technical report, 2022

    Adrien Pavao, Isabelle Guyon, Anne-Catherine Letour- nel, Xavier Bar ´o, Hugo Escalante, Sergio Escalera, Tyler Thomas, and Zhen Xu. Codalab competitions: An open source platform to organize scientific challenges.Technical report, 2022. 2, 3

    2022

  43. [51]

    Film: Visual reasoning with a general conditioning layer

    Ethan Perez, Florian Strub, Harm De Vries, Vincent Du- moulin, and Aaron Courville. Film: Visual reasoning with a general conditioning layer. InProceedings of the AAAI con- ference on artificial intelligence, volume 32, 2018. 8

    2018

  44. [52]

    L. Qu, J. Tian, S. He, Y . Tang, and R. W. H. Lau. Deshad- ownet: A multi-context embedding deep network for shadow removal. In2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 2308–2316, July 2017. 1, 2, 6

    2017

  45. [53]

    Sam 2: Segment anything in images and videos

    Nikhila Ravi, Valentin Gabeur, Yuan-Ting Hu, Ronghang Hu, Chaitanya Ryali, Tengyu Ma, Haitham Khedr, Roman R¨adle, Chloe Rolland, Laura Gustafson, Eric Mintun, Junt- ing Pan, Kalyan Vasudev Alwala, Nicolas Carion, Chao- Yuan Wu, Ross Girshick, Piotr Doll´ar, and Christoph Feic...

    2024 arXiv

  46. [54]

    The tenth NTIRE 2025 efficient super- resolution challenge report

    Bin Ren, Hang Guo, Lei Sun, Zongwei Wu, Radu Timo- fte, Yawei Li, et al. The tenth NTIRE 2025 efficient super- resolution challenge report. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2025. 2

    2025

  47. [55]

    U- net: Convolutional networks for biomedical image segmen- tation

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. U- net: Convolutional networks for biomedical image segmen- tation. InMedical Image Computing and Computer-Assisted Intervention–MICCAI 2015: 18th International Conference, Munich, Germany, October 5-9, 2015, Proceedings, Par...

    2015

  48. [56]

    NTIRE 2025 challenge on UGC video enhancement: Meth- ods and results

    Nickolay Safonov, Alexey Bryntsev, Andrey Moskalenko, Dmitry Kulikov, Dmitriy Vatolin, Radu Timofte, et al. NTIRE 2025 challenge on UGC video enhancement: Meth- ods and results. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Work- sh...

    2025

  49. [57]

    The shadow meets the mask: Pyramid-based shadow removal.Computer Graphics Forum, 27(2):577–586, Apr

    Yael Shor and Dani Lischinski. The shadow meets the mask: Pyramid-based shadow removal.Computer Graphics Forum, 27(2):577–586, Apr. 2008. 1

    2008

  50. [58]

    NTIRE 2025 challenge on event-based image deblurring: Methods and results

    Lei Sun, Andrea Alfarano, Peiqi Duan, Shaolin Su, Kaiwei Wang, Boxin Shi, Radu Timofte, Danda Pani Paudel, Luc Van Gool, et al. NTIRE 2025 challenge on event-based image deblurring: Methods and results. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Re...

    2025

  51. [59]

    The tenth ntire 2025 image denoising challenge report

    Lei Sun, Hang Guo, Bin Ren, Luc Van Gool, Radu Timo- fte, Yawei Li, et al. The tenth ntire 2025 image denoising challenge report. InProceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition (CVPR) Workshops, 2025. 2

    2025

  52. [60]

    Restoring images in adverse weather condi- tions via histogram transformer

    Shangquan Sun, Wenqi Ren, Xinwei Gao, Rui Wang, and Xiaochun Cao. Restoring images in adverse weather condi- tions via histogram transformer. InEuropean Conference on Computer Vision, pages 111–129. Springer, 2024. 7

    2024

  53. [61]

    Shadow removal with paired and un- paired learning

    Florin-Alexandru Vasluianu, Andr ´es Romero, Luc Van Gool, and Radu Timofte. Shadow removal with paired and un- paired learning. InProceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition, pages 826–835, 2021. 1, 2

    2021

  54. [62]

    Wsrd: A novel benchmark for high resolution image shadow removal

    Florin-Alexandru Vasluianu, Tim Seizinger, and Radu Tim- ofte. Wsrd: A novel benchmark for high resolution image shadow removal. InProceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition, 2023. 1, 2, 4, 8

    2023

  55. [63]

    Ntire 2023 im- age shadow removal challenge report

    Florin-Alexandru Vasluianu, Tim Seizinger, Radu Timofte, Shuhao Cui, Junshi Huang, Shuman Tian, Mingyuan Fan, Jiaqi Zhang, Li Zhu, Xiaoming Wei, et al. Ntire 2023 im- age shadow removal challenge report. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern R...

    2023

  56. [64]

    Towards image ambient lighting normalization.arXiv preprint arXiv:2403.18730,

    Florin-Alexandru Vasluianu, Tim Seizinger, Zongwei Wu, Rakesh Ranjan, and Radu Timofte. Towards image ambient lighting normalization.arXiv preprint arXiv:2403.18730,

    arXiv

  57. [65]

    Florin-Alexandru Vasluianu, Tim Seizinger, Zhuyun Zhou, Zongwei Wu, Cailian Chen, Radu Timofte, Wei Dong, Han Zhou, Yuqiong Tian, Jun Chen, Xueyang Fu, Xin Lu, Yurui Zhu, Xi Wang, Dong Li, Jie Xiao, Yunpeng Zhang, Zheng- Jun Zha, Zhao Zhang, Suiyi Zhao, Bo Wang, Yan Luo, Yanya...

    2024

  58. [66]

    NTIRE 2025 ambi- ent lighting normalization challenge

    Florin-Alexandru Vasluianu, Tim Seizinger, Zhuyun Zhou, Zongwei Wu, Radu Timofte, et al. NTIRE 2025 ambi- ent lighting normalization challenge. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2025. 2

    2025

  59. [67]

    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, 40(3):682–695, March 2018. 1

    2018

  60. [68]

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

    Jifeng Wang, Xiang Li, and Jian Yang. Stacked conditional generative adversarial networks for jointly learning shadow detection and shadow removal. InProceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 1788–1797, 2018. 1, 2, 4, 6

    2018

  61. [69]

    NTIRE 2025 challenge on light field image super-resolution: Methods and results

    Yingqian Wang, Zhengyu Liang, Fengyuan Zhang, Lvli Tian, Longguang Wang, Juncheng Li, Jungang Yang, Radu Timofte, Yulan Guo, et al. NTIRE 2025 challenge on light field image super-resolution: Methods and results. InPro- ceedings of the IEEE/CVF Conference on Computer Vision an...

    2025

  62. [70]

    Image quality assessment: from error visibility to structural similarity.IEEE transactions on image processing, 13(4):600–612, 2004

    Zhou Wang, Alan C Bovik, Hamid R Sheikh, and Eero P Si- moncelli. Image quality assessment: from error visibility to structural similarity.IEEE transactions on image processing, 13(4):600–612, 2004. 3

    2004

  63. [71]

    Brown, and Heung-Yeung Shum

    Tai-Pang Wu, Chi-Keung Tang, Michael S. Brown, and Heung-Yeung Shum. Natural shadow matting.ACM Trans. Graph., 26(2):8–es, June 2007. 1

    2007

  64. [72]

    Omnisr: Shadow re- moval under direct and indirect lighting.arXiv preprint arXiv:2410.01719, 2024

    Jiamin Xu, Zelong Li, Yuxin Zheng, Chenyu Huang, Ren- shu Gu, Weiwei Xu, and Gang Xu. Omnisr: Shadow re- moval under direct and indirect lighting.arXiv preprint arXiv:2410.01719, 2024. 5

    2024 arXiv

  65. [73]

    NTIRE 2025 challenge on single image reflection removal in the wild: Datasets, methods and results

    Kangning Yang, Jie Cai, Ling Ouyang, Florin-Alexandru Vasluianu, Radu Timofte, Jiaming Ding, Huiming Sun, Lan Fu, Jinlong Li, Chiu Man Ho, Zibo Meng, et al. NTIRE 2025 challenge on single image reflection removal in the wild: Datasets, methods and results. InProceedings of the...

    2025

  66. [74]

    Towards efficient and scale-robust ultra- high-definition image demoir ´eing

    Xin Yu, Peng Dai, Wenbo Li, Lan Ma, Jiajun Shen, Jia Li, and Xiaojuan Qi. Towards efficient and scale-robust ultra- high-definition image demoir ´eing. InEuropean Conference on Computer Vision, pages 646–662. Springer, 2022. 7

    2022

  67. [75]

    NTIRE 2025 challenge on hr depth from images of specular and transparent surfaces

    Pierluigi Zama Ramirez, Fabio Tosi, Luigi Di Stefano, Radu Timofte, Alex Costanzino, Matteo Poggi, Samuele Salti, Ste- fano Mattoccia, et al. NTIRE 2025 challenge on hr depth from images of specular and transparent surfaces. InPro- ceedings of the IEEE/CVF Conference on Comput...

    2025

  68. [76]

    Restormer: Efficient transformer for high-resolution image restoration

    Waqas Zamir, Syed, Aditya Arora, Salman Khan, Mu- nawar Hayat, Fahad Khan, Shahbaz, and Ming-Hsuang Yan. Restormer: Efficient transformer for high-resolution image restoration. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022. 6

    2022

  69. [77]

    Rgb ↔x: Image decomposition and syn- thesis using material- and lighting-aware diffusion mod- els

    Zheng Zeng, Valentin Deschaintre, Iliyan Georgiev, Yan- nick Hold-Geoffroy, Yiwei Hu, Fujun Luan, Ling-Qi Yan, and Milos Hasan. Rgb ↔x: Image decomposition and syn- thesis using material- and lighting-aware diffusion mod- els. In Andres Burbano, Denis Zorin, and Wojciech Jaros...

    2024

  70. [78]

    The unreasonable effectiveness of deep features as a perceptual metric

    Richard Zhang, Phillip Isola, Alexei A Efros, Eli Shechtman, and Oliver Wang. The unreasonable effectiveness of deep features as a perceptual metric. InCVPR, 2018. 3

    2018

  71. [79]

    Spa- former: Transformer image shadow detection and removal via spatial attention.arXiv e-prints, pages arXiv–2206, 2022

    Xiao Feng Zhang, Chao Chen Gu, and Shan Ying Zhu. Spa- former: Transformer image shadow detection and removal via spatial attention.arXiv e-prints, pages arXiv–2206, 2022. 1

    2022

  72. [80]

    Lita-gs: Illumination- agnostic novel view synthesis via reference-free 3d gaus- sian splatting and physical priors.arXiv preprint arXiv:2504.00219, 2025

    Han Zhou, Wei Dong, and Jun Chen. Lita-gs: Illumination- agnostic novel view synthesis via reference-free 3d gaus- sian splatting and physical priors.arXiv preprint arXiv:2504.00219, 2025. 7

    2025 arXiv

  73. [81]

    Glare: Low light image enhancement via generative latent feature based codebook retrieval

    Han Zhou, Wei Dong, Xiaohong Liu, Shuaicheng Liu, Xiongkuo Min, Guangtao Zhai, and Jun Chen. Glare: Low light image enhancement via generative latent feature based codebook retrieval. InEuropean Conference on Computer Vision, pages 36–54. Springer, 2024. 7

    2024

  74. [82]

    Low-light image enhance- ment via generative perceptual priors.arXiv preprint arXiv:2412.20916, 2024

    Han Zhou, Wei Dong, Xiaohong Liu, Yulun Zhang, Guang- tao Zhai, and Jun Chen. Low-light image enhance- ment via generative perceptual priors.arXiv preprint arXiv:2412.20916, 2024. 7

    2024 arXiv

  75. [83]

    Breaking through the haze: An advanced non-homogeneous dehazing method based on fast fourier convolution and convnext

    Han Zhou, Wei Dong, Yangyi Liu, and Jun Chen. Breaking through the haze: An advanced non-homogeneous dehazing method based on fast fourier convolution and convnext. In Proceedings of the IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition Workshops, 2023. 5, 6, 7

    2023

  76. [84]

    Srformer: Permuted self-attention for single image super-resolution

    Yupeng Zhou, Zhen Li, Chun-Le Guo, Song Bai, Ming-Ming Cheng, and Qibin Hou. Srformer: Permuted self-attention for single image super-resolution. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 12780–12791, 2023. 8

    2023

  77. [85]

    Hue guidance network for single image reflection removal.IEEE Transactions on Neural Net- works and Learning Systems, 35(10):13701–13712, 2024

    Yurui Zhu, Xueyang Fu, Zheyu Zhang, Aiping Liu, Zhiwei Xiong, and Zheng-Jun Zha. Hue guidance network for single image reflection removal.IEEE Transactions on Neural Net- works and Learning Systems, 35(10):13701–13712, 2024. 8

    2024

  78. [86]

    En- hanced coarse-to-fine network for image restoration from under-display cameras

    Yurui Zhu, Xi Wang, Xueyang Fu, and Xiaowei Hu. En- hanced coarse-to-fine network for image restoration from under-display cameras. InComputer Vision – ECCV 2022 Workshops: Tel Aviv, Israel, October 23–27, 2022, Pro- ceedings, Part V, page 130–146, Berlin, Heidelberg, 2023. Sp...

    2022

Pith tools

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