Pith. sign in

REVIEW 5 major objections 5 minor 2 cited by

Bayesian Neural Networks for One-to-Many Mapping in Image Enhancement

T0 review · 5 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read A Bayesian network that samples weights from a learned posterior maps one degraded input to many plausible enhanced images, and a two-stage design delivers state-of-the-art low-light and underwater enhancement fast.

desk verdict A genuinely new BNN-based take on a real one-to-many problem; the MC baseline is solid but the headline SOTA is oracle-ranked and the central multi-reference claim is untested. read the letter →

arxiv 2501.14265 v3 pith:IVXA6EQQ submitted 2025-01-24 cs.CV

Guoxi Huang , Qirui Yang , Ruirui Lin , Zipeng Qi , David Bull , Nantheera Anantrasirichai This is my paper · ORCID
classification cs.CV
keywords Bayesianneuralnetworksone-to-manymappinglow-lightimageenhancementunderwatervariationalinferenceuncertaintymodelingtwo-stageframework
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 low-light and underwater enhancement is inherently a one-to-many problem — one degraded input has many plausible enhanced targets — and that deterministic networks, which learn a single mapping, are the wrong tool for it. The proposed Bayesian Enhancement Model (BEM) instead treats network weights as random variables drawn from a learned variational posterior, so each forward pass returns a different plausible output. To keep repeated sampling affordable, BEM splits the work: a Bayesian network samples many coarse illumination maps in a low-dimensional space, a score-based or Monte Carlo step picks one, and a deterministic network refines it into full-resolution detail. The paper reports that this design beats deterministic and probabilistic baselines on the LOL and UIEB benchmarks while running about 22 times faster than a full-resolution Bayesian network and on par with a plain deterministic network.

What carries the argument

The load-bearing machinery is the reparameterized Bayesian weight posterior. Every layer of the Stage I backbone becomes Bayesian, with weights sampled as $w = \mu + \sigma \odot \epsilon$, $\epsilon \sim \mathcal{N}(0,I)$, so $K$ forward passes yield $K$ distinct coarse outputs; an Adaptive Prior, updated by exponential moving average from the posterior parameters, stabilizes training and speeds convergence. The second load-bearing piece is the two-stage decomposition itself: the BNN works on a low-pass-filtered, downsampled input, and the Stage II refinement network is trained not on the BNN's own latents but on the analytic pseudo-ground-truth $z = \mathrm{LP}\big((\sqrt{x^2 + 4\alpha y} - x)/(2\alpha)\big)$, obtained by solving $\tilde{y} = (x + \alpha z) \odot z$ for $z$ with $\tilde{y}$ set to the true target $y$. This substitution is what prevents the deterministic stage from regressing the BNN's diverse samples back into a single mode.

What would settle it

Feed the Stage II refinement network the true inference-time latents from Eq. (7) instead of the analytic pseudo-ground-truth latents on the same paired benchmarks, and compare output quality; equivalently, measure a distributional distance such as FID or MMD between the two latent sets. If quality drops markedly or the distributions diverge, the premise that the two-stage pipeline refines what Stage I actually produces is unsupported.

Watch

Extended reading notes

Core claim

The central discovery claim is that weight-space Bayesian sampling is a practical way to model ambiguity in image enhancement, and that the ambiguity can be resolved cheaply by separating coarse diversity from fine detail. Concretely, Stage I passes a low-pass-filtered, downsampled input through a Bayesian backbone and draws $K$ weight samples $w = \mu + \sigma \odot \epsilon$, producing $K$ coarse latent illumination maps $\{z_k\}$; each candidate is scored through the closed-form proxy $\tilde{y} = (x + \alpha z) \odot z$, and the best-scoring (or the averaged) latent $z^*$ is handed to a deterministic Stage II network that writes the high-frequency detail. The paper shows the same recipe works with Transformer and Mamba backbones, that it improves PSNR, SSIM, and LPIPS simultaneously on LOL-v1/v2 and UIEB-R90 against deterministic baselines (KinD, Restormer, RetinexFormer, SNR-Net) and probabilistic ones (LLFlow, GLARE, diffusion, VAE), and that its pixel-wise output variability is far larger than a VAE's or a diffusion model's while remaining low along object boundaries. The authors report the ranking-based variant as an upper bound in full-reference tables, because ranking can in effect select toward the reference image; the Monte Carlo variant is the directly comparable number and is still competitive or better.

Load-bearing premise

The load-bearing assumption is that the coarse latent $z$ the Stage I network actually produces at inference — an upsampled, stochastically sampled map — lies on the same manifold as the analytic pseudo-ground-truth $z$ used to train Stage II, and the paper reports no analysis or ablation comparing these two distributions.

Editorial extensions

If this is right

  • One-to-many modeling can replace one-to-one regression as the default framing for low-light and underwater enhancement: the same two-stage recipe improves PSNR, SSIM, and LPIPS at the same time on LOL-v1, LOL-v2, and UIEB-R90.
  • Coarse-space Bayesian sampling is enough to capture the diversity, so uncertainty modeling does not have to cost full-resolution latency: BEM is reported to run about $22\times$ faster than a full-resolution BNN and about as fast as a plain DNN on $512\times512$ images.
  • The framework is backbone-agnostic: both the Transformer and Mamba instantiations deliver the reported gains, so the design inherits future backbone advances.
  • The two inference modes are complementary tools: ranking-based inference ($\text{BEM}_{\text{Rank}}$) produces sharper, higher-contrast results and best no-reference scores on C60, U45, and five unpaired low-light sets, while Monte Carlo averaging ($\text{BEM}_{\text{MC}}$) yields more conservative, noise-suppressed outputs.
  • The spread of BEM's repeated predictions carries signal, not just noise: score distributions show a low-scoring tail consistent with noisy reference labels, supporting the claim that weight-level uncertainty absorbs label noise instead of committing to one arbitrary target.

Reading between the lines

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

  • The paper trains Stage II on the analytic pseudo-ground-truth latent but runs it on the BNN's own upsampled stochastic latents; a natural testable extension is to train Stage II directly on latents drawn from the real inference distribution and measure whether the reported gains persist or improve.
  • Because the ranking mode is steered by a user-selected no-reference metric (CLIP-IQA, NIQE, UIQM, UCIQE), the pipeline doubles as a tunable aesthetic prior: the same trained model could be pushed toward different notions of quality at inference time without retraining, a use the paper only begins to illustrate.
  • The same BNN-DNN decomposition should transfer to other one-to-many restoration problems such as deblurring, deraining, or super-resolution with unknown degradations, where the coarse stage would need to encode a different ambiguity structure than illumination; the paper does not test this.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 5 minor

Summary. The paper proposes the Bayesian Enhancement Model (BEM), a two-stage BNN-DNN architecture for low-light and underwater image enhancement. Stage I uses a Bayesian neural network to map a downsampled, low-pass-filtered input to a coarse latent illumination map, sampling weights from a variational posterior; Stage II uses a deterministic DNN to refine fine details given the input and the coarse latent. Inference can either rank a set of latent candidates with a no-reference image quality metric (BEMRank) or average the candidates (BEMMC). The authors report full-reference results on LOL-v1/v2 and UIEB-R90, no-reference results on LIME, NPE, MEF, DICM, VV, C60, and U45, and qualitative and quantitative analyses of output diversity.

Significance. The problem of one-to-many mapping in image enhancement is real and underexplored, and the two-stage BNN-DNN design is a sensible attempt to make Bayesian inference practical for high-resolution images. The paper ships code and evaluates on multiple established benchmarks, which is a strength. The adaptive prior and the ranking-based selection are interesting ideas. However, the central claim that BEM models a true one-to-many target distribution is not directly demonstrated, and the reported state-of-the-art results partly rely on oracle candidate selection that is not a deployable inference procedure. If the one-to-many claim can be supported with multi-reference evaluation and the actual (non-oracle) inference variants are shown to be robust, the contribution would be valuable.

major comments (5)
  1. [Sec. 4.1, Tables 1-2] The full-reference state-of-the-art claim is based on the BEMRank rows, which the paper itself marks in gray as an upper bound and 'not directly comparable to the other results.' The sentence 'our BEM ... achieves superior performance across all metrics and datasets' therefore conflates an oracle selection procedure with the actual method. The deployable MC variant is competitive but not uniformly dominant; for example, on LOL-v1 BEM-MC has PSNR 28.30 vs. 27.84 for GlobalDiff, a small margin, and on LOL-v2-real it trails GLARE in PSNR (31.41 vs. 28.98? actually BEM-MC is higher, but the margins are not consistent across all metrics). At minimum, the abstract and Section 4.1 should state that the upper-bound rows are not part of the SOTA claim, and the actual inference modes should be compared separately with error bars.
  2. [Sec. 3.1, Eq. (6)] The central one-to-many claim is not supported by the training objective or the evaluation. Eq. (6) is a single-target regression loss (L2 plus KL) over paired data with one reference per input; it contains no term that encourages the sampled outputs to cover multiple plausible reference images. Figures 8-10 demonstrate output variability, but variability alone does not establish that the samples lie on the manifold of plausible enhancement targets or that they cover the true ambiguity. Without a multi-reference dataset or a task-specific coverage evaluation, the paper has not shown that the BNN posterior corresponds to one-to-many target structure rather than variational noise or label noise.
  3. [Sec. 3.2, Eqs. (7)-(9)] There is a train/inference mismatch for the coarse latent z. During Stage II training, z is replaced by the analytic solution LP((sqrt(x^2+4αy)-x)/(2α)), but at inference z comes from the BNN in Eq. (7), namely Up(F(Down(LP(x),r);w)), or from the upsampled candidate in Algorithm 1. These two quantities differ in how they are filtered, downsampled, and upsampled, and the paper provides no analysis, visualization, or ablation showing that the inference-time z distribution matches the training-time pseudo-ground-truth distribution. If they are mismatched, the Stage II DNN is trained on inputs it will not encounter at test time, which would directly degrade output quality.
  4. [Sec. 3.1, 'Adaptive Prior'] The adaptive prior is a claimed contribution, but no experiment isolates its effect. The text says it 'has been shown to achieve faster convergence than fixed or empirical priors,' yet no citation or ablation is provided. Since the prior is defined as an EMA of the posterior parameters, it is self-referential and could in principle collapse the KL term; the paper should include an ablation with and without the adaptive prior, and with fixed priors, to support the claim.
  5. [Tables 1-3] No error bars or multiple-seed results are reported. Because both BEMRank and BEMMC are stochastic, point estimates from a single run are insufficient to establish significant improvements over deterministic baselines, especially where margins are small. Reporting means and standard deviations over at least three runs, or paired significance tests, would materially strengthen the empirical claims.
minor comments (5)
  1. [Sec. 3.3] The description of BEMRank in the full-reference tables is confusing: Algorithm 1 ranks by a no-reference IQA, but Table 1 states that BEMRank selects the candidate closest to the reference under the chosen metric. The paper should clarify which ranking is used in which table and distinguish 'oracle rank' from 'IQA rank' clearly.
  2. [Sec. 4.1, Table 3] For the no-reference evaluation, BEMRank ranks candidates with NIQE and then reports NIQE on the selected candidate. This is not circular as a method comparison, but it is a form of test-time optimization; the paper should state this explicitly and also report results with a ranking metric different from the evaluation metric.
  3. [Sec. 3.2] Equation (8), tilde-y = (x + α z) ⊙ z, is introduced as a modeling choice with only informal motivation. A brief derivation or a toy experiment justifying why this composition is preferable to x+z or x⊙z would improve reproducibility.
  4. [References] There are several formatting issues in the references and text, including 'V onikakis' (missing letter), duplicated method names in Figure 7's caption, and inconsistent use of 'VAEs' spacing. These should be cleaned up.
  5. [Fig. 2] The figure does not clearly show the resolution at which z is produced and where the upsampling happens relative to the ranking operation. Aligning the figure with Algorithm 1 would help readers understand the pipeline.

Circularity Check

1 steps flagged · score 5.0 of 10

Full-reference BEMRank scores are oracle-selected upper bounds, making part of the SOTA claim reduce to the selection rule; the core BNN formulation and BEMMC results remain non-circular.

  1. fitted input called prediction [Sec. 4.1 (Evaluation), full-reference comparisons on LOL-v1/v2 and UIEB-R90 (Tables 1-2)]
    "For full-reference evaluation, we report BEM Rank as an upper-bound performance of our method, since its first-stage ranking selects the candidate that is closest to the reference image under the chosen metric."

    The reported PSNR/SSIM/LPIPS values for BEMRank are not outputs of a fixed learned inference rule; they are obtained by selecting, among K stochastic samples, the one closest to the reference under the same metric being reported. This makes the score an oracle maximum over the sample distribution by construction: any stochastic model that emits diverse candidates would receive the same upper-bound score under this selection rule. The surrounding text ('BEM...achieves superior performance across all metrics and datasets') presents these oracle-selected values in the main comparison, so part of the benchmark claim reduces to the selection procedure rather than to the learned BNN posterior.

full rationale

The only concrete reduction I found is the oracle-selected BEMRank full-reference evaluation. The paper's central methodological contribution--using a BNN to sample weights and a two-stage BNN-DNN refinement--is not itself equivalent to its inputs: Stage I is trained with an ELBO-like loss, Stage II is trained with an L1 loss against the analytic coarse latent, and the benchmarks are external data. The adaptive prior is self-referential in that the prior is updated from the posterior EMA, but it is presented as a training heuristic and no theorem or headline result is derived from it, so it does not constitute load-bearing circularity. The lack of multi-reference ground truth means the claim that the sampled outputs cover the true one-to-many target distribution is under-supported (variability is not coverage), but that is an empirical-evidence gap rather than a circular derivation, so I do not count it as a circular step. Self-citations (Huang et al. 2025; Malyugina et al. 2025) appear only in related-work lists and are not load-bearing. Overall the core derivation is self-contained; the score is elevated only because part of the headline SOTA claim uses an oracle selection rule whose full-reference metrics are forced by construction.

Assumptions & free parameters 4 free parameters · 6 assumptions · 0 invented entities

The method introduces no new physical entities. Its contribution rests on standard variational inference plus hand-chosen hyperparameters and an unvalidated composition model.

free parameters (4)
  • α (alpha) = 0.025
    Scalar in Eq. (8) blending input x and coarse illumination z. Chosen by hand; no sensitivity analysis.
  • downscale factor r = 1/16
    Downsampling ratio for Stage I; chosen by hand.
  • number of candidates K = 25
    Number of weight samples or ranked candidates; chosen by hand.
  • EMA decay rate for adaptive prior = not specified
    The adaptive prior updates via EMA, but the decay rate is not reported.
assumptions (6)
  • domain assumption Pixel-wise prediction errors follow a Gaussian distribution, giving an L2 data term.
    Sec. 2.1 models P(y|x,w) as a multivariate Gaussian with mean F(x;w); this justifies the squared error in Eq. (6).
  • standard math The variational posterior is a diagonal Gaussian, and weights are sampled via the reparameterization trick.
    Sec. 3.1, Eq. (5); standard Bayes-by-backprop.
  • ad hoc to paper The prior P(w) can be defined adaptively as an EMA of the posterior mean and std.
    Sec. 3.1, after Eq. (5); no theoretical justification or ablation.
  • ad hoc to paper The composition tilde-y = (x + α z) ⊙ z is a valid model of how the coarse illumination z and input x produce the enhanced image.
    Sec. 3.2, Eq. (8); a modeling choice without derivation from physics.
  • domain assumption Low-pass filtering and downsampling to resolution r preserves the one-to-many information needed for uncertainty modeling.
    Sec. 3.2, Eq. (7); no experiments analyzing information loss.
  • domain assumption A no-reference IQA metric (e.g., CLIP-IQA, NIQE) can select the perceptually best candidate among the BNN samples.
    Sec. 3.3; the ranking relies on an external metric's correlation with human preference.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Bayesian Neural Networks for One-to-Many Mapping in Image Enhancement." pith.science (2026). https://pith.science/paper/IVXA6EQQ

@misc{pith2026250114265,
  author       = {Pith},
  title        = {Pith review of: Bayesian Neural Networks for One-to-Many Mapping in Image Enhancement},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/IVXA6EQQ}},
  note         = {Machine review of arXiv:2501.14265}
}
read the original abstract

In image enhancement tasks, such as low-light and underwater image enhancement, a degraded image can correspond to multiple plausible target images due to dynamic photography conditions. This naturally results in a one-to-many mapping problem. To address this, we propose a Bayesian Enhancement Model (BEM) that incorporates Bayesian Neural Networks (BNNs) to capture data uncertainty and produce diverse outputs. To enable fast inference, we introduce a BNN-DNN framework: a BNN is first employed to model the one-to-many mapping in a low-dimensional space, followed by a Deterministic Neural Network (DNN) that refines fine-grained image details. Extensive experiments on multiple low-light and underwater image enhancement benchmarks demonstrate the effectiveness of our method.

Figures

Figures reproduced from arXiv: 2501.14265 by the authors.

Figure 1
Figure 1. One-to-Many Mapping where an image crop x as￾sociated with multiple targets {y 1 , . . . , y 6}. A DNN (left) can only predict one of the targets. In contrast, a BNN (right) can produce many predictions according to a learned proba￾bility distribution. into the training data—i.e., label noise. Such label noise is further amplified in extremely low-visibility data collec￾tion environments—particularly in challenging … view at source ↗
Figure 2
Figure 2. The two-stage pipeline. In Stage I, the BNN with weights [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Visual comparisons of the DNN baseline, BEMMC, and BEMRank with CLIP-IQA. The rightmost patches highlight [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: One-to-many mapping from input to outputs. The [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Inference speed on an Nvidia RTX 4090. primary concern. Inference Speed. Algorithm 1 avoids redundant sam￾pling, resulting in a substantial reduction in inference la￾tency [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]
Figure 7
Figure 7. Figure 7: Visual comparisons on the R90, C60 and U45 datasets. Best viewed when zoomed in. [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]
Figure 8
Figure 8. Figure 8: Score distributions of 500 predictions from BEM [PITH_FULL_IMAGE:figures/full_fig_p007_8.png]
Figure 9
Figure 9. Figure 9: Visualization of pixel-wise output variability from [PITH_FULL_IMAGE:figures/full_fig_p007_9.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

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

  1. Marine Snow Removal Using Internally Generated Pseudo Ground Truth

    cs.CV 2025-04 conditional novelty 5.0 of 10

    A dataset built from real marine snow masks extracted by median frame subtraction enables a video enhancement network to reduce snow and improve SLAM feature matching and dense mapping.

  2. Visual enhancement and 3D representation for underwater scenes: a review

    cs.CV 2025-05 conditional novelty 2.0 of 10

    A structured review of underwater visual enhancement and 3D reconstruction methods, from classical physics-based approaches to NeRF and 3D Gaussian Splatting, with a small qualitative comparison.

Reference graph

Works this paper leans on

56 extracted references · 37 canonical work pages · cited by 2 Pith papers

  1. [1]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.a...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

  3. [3]

    Bai, J.; Yin, Y.; and He, Q. 2024. Retinexmamba: Retinex-based Mamba for Low-light Image Enhancement. arXiv preprint arXiv:2405.03349

    arXiv 2024

  4. [4]

    Blundell, C.; Cornebise, J.; Kavukcuoglu, K.; and Wierstra, D. 2015. Weight uncertainty in neural network. In International conference on machine learning, 1613--1622. PMLR

    work page 2015

  5. [5]

    Cai, Y.; Bian, H.; Lin, J.; Wang, H.; Timofte, R.; and Zhang, Y. 2023. Retinexformer: One-stage retinex-based transformer for low-light image enhancement. In Proceedings of the IEEE/CVF International Conference on Computer Vision, 12504--12513

    work page 2023

  6. [6]

    Cong, R.; Yang, W.; Zhang, W.; Li, C.; Guo, C.-L.; Huang, Q.; and Kwong, S. 2023. Pugan: Physical model-guided underwater image enhancement using gan with dual-discriminators. IEEE Transactions on Image Processing, 32: 4472--4485

    work page 2023

  7. [7]

    Dusenberry, M.; Jerfel, G.; Wen, Y.; Ma, Y.; Snoek, J.; Heller, K.; Lakshminarayanan, B.; and Tran, D. 2020. Efficient and scalable bayesian neural nets with rank-1 factors. In International conference on machine learning, 2782--2792. PMLR

    work page 2020

  8. [8]

    Fu, H.; Zheng, W.; Meng, X.; Wang, X.; Wang, C.; and Ma, H. 2023 a . You do not need additional priors or regularizers in retinex-based low-light image enhancement. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 18125--18134

    work page 2023

Show all 56 references
  1. [9]

    Fu, Z.; Wang, W.; Huang, Y.; Ding, X.; and Ma, K.-K. 2022. Uncertainty inspired underwater image enhancement. In European conference on computer vision, 465--482. Springer

    2022

  2. [10]

    Fu, Z.; Yang, Y.; Tu, X.; Huang, Y.; Ding, X.; and Ma, K.-K. 2023 b . Learning a simple low-light image enhancer from paired low-light instances. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 22252--22261

    2023

  3. [11]

    Gal, Y.; and Ghahramani, Z. 2016. Dropout as a bayesian approximation: Representing model uncertainty in deep learning. In international conference on machine learning, 1050--1059. PMLR

    2016

  4. [12]

    Gu, A.; and Dao, T. 2023. Mamba: Linear-time sequence modeling with selective state spaces. arXiv preprint arXiv:2312.00752

    2023 arXiv

  5. [13]

    C.; Hou, J.; Kwong, S.; and Cong, R

    Guo, C.; Li, C.; Guo, J.; Loy, C. C.; Hou, J.; Kwong, S.; and Cong, R. 2020. Zero-reference deep curve estimation for low-light image enhancement. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 1780--1789

    2020

  6. [14]

    Guo, X.; Li, Y.; and Ling, H. 2016. LIME: Low-light image enhancement via illumination map estimation. IEEE Transactions on image processing, 26(2): 982--993

    2016

  7. [15]

    Hai, J.; Xuan, Z.; Yang, R.; Hao, Y.; Zou, F.; Lin, F.; and Han, S. 2023. R2rnet: Low-light image enhancement via real-low to real-normal network. Journal of Visual Communication and Image Representation, 90: 103712

    2023

  8. [16]

    Han, J.; Shoeiby, M.; Malthus, T.; Botha, E.; Anstee, J.; Anwar, S.; Wei, R.; Petersson, L.; and Armin, M. A. 2021. Single underwater image restoration by contrastive learning. In 2021 IEEE international geoscience and remote sensing symposium IGARSS, 2385--2388. IEEE

    2021

  9. [17]

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

    2020

  10. [18]

    E.; and Van Camp, D

    Hinton, G. E.; and Van Camp, D. 1993. Keeping the neural networks simple by minimizing the description length of the weights. In Proceedings of the sixth annual conference on Computational learning theory, 5--13

    1993

  11. [19]

    Hou, J.; Zhu, Z.; Hou, J.; Liu, H.; Zeng, H.; and Yuan, H. 2024. Global structure-aware diffusion process for low-light image enhancement. Advances in Neural Information Processing Systems, 36

    2024

  12. [20]

    Huang, G.; Lin, R.; Li, Y.; Bull, D.; and Anantrasirichai, N. 2025. BVI-Mamba: video enhancement using a visual state-space model for low-light and underwater environments. In Machine Learning from Challenging Data 2025, volume 13460, 74--81. SPIE

    2025

  13. [21]

    Huang, S.; Wang, K.; Liu, H.; Chen, J.; and Li, Y. 2023. Contrastive semi-supervised learning for underwater image restoration via reliable bank. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 18145--18155

    2023

  14. [22]

    Huo, F.; Li, B.; and Zhu, X. 2021. Efficient wavelet boost learning-based multi-stage progressive refinement network for underwater image enhancement. In Proceedings of the IEEE/CVF international conference on computer vision, 1944--1952

    2021

  15. [23]

    J.; Xia, Y.; and Sattar, J

    Islam, M. J.; Xia, Y.; and Sattar, J. 2020. Fast underwater image enhancement for improved visual perception. IEEE Robotics and Automation Letters, 5(2): 3227--3234

    2020

  16. [24]

    Jiang, J.; Ye, T.; Bai, J.; Chen, S.; Chai, W.; Jun, S.; Liu, Y.; and Chen, E. 2023. Five A ^ + Network: You Only Need 9K Parameters for Underwater Image Enhancement. British Machine Vision Conference (BMVC)

    2023

  17. [25]

    Jiang, Y.; Gong, X.; Liu, D.; Cheng, Y.; Fang, C.; Shen, X.; Yang, J.; Zhou, P.; and Wang, Z. 2021. Enlightengan: Deep light enhancement without paired supervision. IEEE transactions on image processing, 30: 2340--2349

    2021

  18. [26]

    Kendall, A.; and Cipolla, R. 2016. Modelling uncertainty in deep learning for camera relocalization. In 2016 IEEE international conference on Robotics and Automation (ICRA), 4762--4769. IEEE

    2016

  19. [27]

    Kendall, A.; Gal, Y.; and Cipolla, R. 2018. Multi-task learning using uncertainty to weigh losses for scene geometry and semantics. In Proceedings of the IEEE conference on computer vision and pattern recognition, 7482--7491

    2018

  20. [28]

    Kingma, D. P. 2014. Auto-encoding variational bayes. International Conference on Learning Representations (ICLR)

    2014

  21. [29]

    Krishnan, R.; Subedar, M.; and Tickoo, O. 2020. Specifying weight priors in bayesian deep neural networks with empirical bayes. In Proceedings of the AAAI conference on artificial intelligence, volume 34, 4477--4484

    2020

  22. [30]

    Lee, C.; Lee, C.; and Kim, C.-S. 2013. Contrast enhancement based on layered difference representation of 2D histograms. IEEE transactions on image processing, 22(12): 5372--5384

    2013

  23. [31]

    Li, C.; Anwar, S.; Hou, J.; Cong, R.; Guo, C.; and Ren, W. 2021. Underwater image enhancement via medium transmission-guided multi-color space embedding. IEEE Transactions on Image Processing, 30: 4985--5000

    2021

  24. [32]

    Li, C.; Guo, C.; Ren, W.; Cong, R.; Hou, J.; Kwong, S.; and Tao, D. 2019. An underwater image enhancement benchmark dataset and beyond. IEEE transactions on image processing, 29: 4376--4389

    2019

  25. [33]

    Li, H.; Li, J.; and Wang, W. 2019. A fusion adversarial underwater image enhancement network with a public test dataset. arXiv preprint arXiv:1906.06819

    2019 arXiv

  26. [34]

    Li, K.; Wu, L.; Qi, Q.; Liu, W.; Gao, X.; Zhou, L.; and Song, D. 2023. Beyond single reference for training: Underwater image enhancement via comparative learning. IEEE Transactions on Circuits and Systems for Video Technology, 33(6): 2561--2576

    2023

  27. [35]

    Liu, R.; Ma, L.; Zhang, J.; Fan, X.; and Luo, Z. 2021. Retinex-inspired unrolling with cooperative prior architecture search for low-light image enhancement. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 10561--10570

    2021

  28. [36]

    Liu, S.; Li, K.; and Ding, Y. 2024. Underwater Image Enhancement by Diffusion Model with Customized CLIP-Classifier. arXiv preprint arXiv:2405.16214

    2024 arXiv

  29. [37]

    Ma, K.; Zeng, K.; and Wang, Z. 2015. Perceptual quality assessment for multi-exposure image fusion. IEEE Transactions on Image Processing, 24(11): 3345--3356

    2015

  30. [38]

    Malyugina, A.; Huang, G.; Ruiz, E.; Leslie, B.; and Anantrasirichai, N. 2025. Marine Snow Removal Using Internally Generated Pseudo Ground Truth. arXiv preprint arXiv:2504.19289

    2025 arXiv

  31. [39]

    Neal, R. M. 2012. Bayesian learning for neural networks, volume 118. Springer Science & Business Media

    2012

  32. [40]

    Peng, L.; Zhu, C.; and Bian, L. 2023. U-shape transformer for underwater image enhancement. IEEE Transactions on Image Processing

    2023

  33. [41]

    Peng, Y.-T.; Cao, K.; and Cosman, P. C. 2018. Generalization of the dark channel prior for single image restoration. IEEE Transactions on Image Processing, 27(6): 2856--2868

    2018

  34. [42]

    Shen, Z.; Xu, H.; Luo, T.; Song, Y.; and He, Z. 2023. UDAformer: Underwater image enhancement based on dual attention transformer. Computers & Graphics, 111: 77--88

    2023

  35. [43]

    Tang, Y.; Kawasaki, H.; and Iwaguchi, T. 2023. Underwater image enhancement by transformer-based diffusion model with non-uniform sampling for skip strategy. In Proceedings of the 31st ACM International Conference on Multimedia, 5419--5427

    2023

  36. [44]

    Tomczak, M.; Swaroop, S.; Foong, A.; and Turner, R. 2021. Collapsed variational bounds for Bayesian neural networks. Advances in Neural Information Processing Systems, 34: 25412--25426

    2021

  37. [45]

    N.; Kaiser, .; and Polosukhin, I

    Vaswani, A.; Shazeer, N.; Parmar, N.; Uszkoreit, J.; Jones, L.; Gomez, A. N.; Kaiser, .; and Polosukhin, I. 2017. Attention is all you need. Advances in neural information processing systems, 30

    2017

  38. [46]

    Vonikakis, V.; Kouskouridas, R.; and Gasteratos, A. 2018. On the evaluation of illumination compensation algorithms. Multimedia Tools and Applications, 77: 9211--9231

    2018

  39. [47]

    Wang, S.; Zheng, J.; Hu, H.-M.; and Li, B. 2013. Naturalness preserved enhancement algorithm for non-uniform illumination images. IEEE transactions on image processing, 22(9): 3538--3548

    2013

  40. [48]

    Wang, Y.; Wan, R.; Yang, W.; Li, H.; Chau, L.-P.; and Kot, A. 2022. Low-light image enhancement with normalizing flow. In Proceedings of the AAAI conference on artificial intelligence, volume 36, 2604--2612

    2022

  41. [49]

    Wei, C.; Wang, W.; Yang, W.; and Liu, J. 2018. Deep retinex decomposition for low-light enhancement. British Machine Vision Conference (BMVC)

    2018

  42. [50]

    Xu, X.; Wang, R.; Fu, C.-W.; and Jia, J. 2022. SNR-aware low-light image enhancement. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, 17714--17724

    2022

  43. [51]

    Yan, Q.; Feng, Y.; Zhang, C.; Pang, G.; Shi, K.; Wu, P.; Dong, W.; Sun, J.; and Zhang, Y. 2025. Hvi: A new color space for low-light image enhancement. In Proceedings of the Computer Vision and Pattern Recognition Conference, 5678--5687

    2025

  44. [52]

    Yang, W.; Wang, W.; Huang, H.; Wang, S.; and Liu, J. 2021. Sparse gradient regularized deep retinex network for robust low-light image enhancement. IEEE Transactions on Image Processing, 30: 2072--2086

    2021

  45. [53]

    W.; Arora, A.; Khan, S.; Hayat, M.; Khan, F

    Zamir, S. W.; Arora, A.; Khan, S.; Hayat, M.; Khan, F. S.; and Yang, M.-H. 2022. Restormer: Efficient transformer for high-resolution image restoration. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 5728--5739

    2022

  46. [54]

    Zhang, W.; Zhuang, P.; Sun, H.-H.; Li, G.; Kwong, S.; and Li, C. 2022. Underwater image enhancement via minimal color loss and locally adaptive contrast enhancement. IEEE Transactions on Image Processing, 31: 3997--4010

    2022

  47. [55]

    Zhang, Y.; Zhang, J.; and Guo, X. 2019. Kindling the darkness: A practical low-light image enhancer. In Proceedings of the 27th ACM international conference on multimedia, 1632--1640

    2019

  48. [56]

    Zhou, H.; Dong, W.; Liu, X.; Liu, S.; Min, X.; Zhai, G.; and Chen, J. 2024. GLARE: Low Light Image Enhancement via Generative Latent Feature based Codebook Retrieval. Proceedings of the European conference on computer vision (ECCV)

    2024

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.