Rethinking Image Histogram Matching for Image Classification

Atsushi Ito; Rikuto Otsuka; Takahiro Toizumi; Yuho Shoji; Yuka Ogino

REVIEW 3 major objections 3 minor 31 references

Rethinking Image Histogram Matching for Image Classification

T0 review · 3 major / 3 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read The paper claims that a single target pixel-value distribution, optimized by the downstream classifier's loss on normal-weather images, generalizes to unseen adverse weather conditions and outperforms conventional histogram-based…

desk verdict A clever, simple learnable histogram target distribution that likely overstates its headline because the key hyperparameter was chosen on the test sets. read the letter →

arxiv 2506.01346 v1 pith:LDKN456M submitted 2025-06-02 cs.CV

Rikuto Otsuka , Yuho Shoji , Yuka Ogino , Takahiro Toizumi , Atsushi Ito This is my paper

classification cs.CV

keywords imagehistogrammatchingdifferentiablepreprocessingadverseweathergeneralizationtargetdistributionlearningclassificationlow-contrastimagesequalizationend-to-endtraining

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 standard choice of a uniform target distribution in histogram matching is not optimal for classifiers, and that a single learned target distribution can serve as a universal preprocessing for images captured under any adverse weather condition. The authors redesign histogram matching so that the target distribution is a set of trainable parameters, optimized end-to-end with the downstream classifier's loss on normal daytime images. At test time, every input image, whatever its weather, is remapped to this fixed learned distribution. Reported numbers: adverse-weather mean top-1 accuracy rises to 52.53% with the proposed method, compared with 40.84% for histogram equalization, 46.84% for CLAHE, and 35.12% for no preprocessing.

What carries the argument

The central object is a differentiable, parametric histogram matcher composed of two parts. The distribution matcher flattens each color channel, sorts pixels by intensity, and replaces the sorted vector with an upsampled version of a trainable parameter vector $P_c \in \mathbb{R}^s$ using linear interpolation; the rank ordering of pixels is preserved. The parameter container holds these trainable values, which define the target distribution. Because the sorting operation is a fixed permutation and the upsampling is linear, the whole transform is differentiable with respect to $P_c$, so the target distribution can be optimized by the cross-entropy loss of the downstream classifier. The learned distribution is shared across all images and weather conditions, and it is what remains after the transform discards absolute intensities and spatial layout.

What would settle it

Train the same pipeline but replace the sorted-index replacement with a fixed random permutation of the learned target values across pixels, so the rank ordering is destroyed; if the accuracy on the adverse-weather test sets stays high, the rank-preservation assumption is not load-bearing, and if it collapses, the method indeed works through rank ordering. A second check is to apply the trained preprocessing to a synthetic dataset where class labels are determined by absolute brightness (for example, bright versus dark objects under constant illumination); failure there while histogram equalization succeeds would indicate that absolute intensities were doing the work.

Watch

Extended reading notes

Core claim

The central claim is that the shape of the target pixel value distribution in histogram matching is itself a learnable object that should be optimized for the downstream classifier, rather than fixed to a uniform distribution. The proposed method sorts each color channel's pixels, then replaces the sorted values with quantiles drawn from a trainable parameter vector, producing an output image whose pixel distribution equals the learned target. This operation is differentiable, so the target distribution and the classifier weights can be trained jointly using cross-entropy on normal-weather images only. The paper reports that the resulting single distribution transfers to fog, rain, sand, snow, and night conditions, outperforming conventional histogram-based preprocessing at the expense of a small drop in daytime accuracy and a visible loss of visual quality.

Load-bearing premise

The method assumes that sorting a color channel by intensity and replacing absolute pixel values with learned quantiles keeps the information that a classifier needs to distinguish classes; if class-discriminative cues live in absolute intensities or in spatial structure rather than in the order of pixel values, the learned target distribution cannot transfer to unseen domains.

Editorial extensions

If this is right

Histogram matching can be repurposed as a trainable layer, and the uniform distribution used by traditional histogram equalization is not the best target for classification.
A preprocessing module can be trained on clean daytime images only and still improve accuracy on unseen adverse-weather domains, with gains of roughly 12 points over histogram equalization and 6 points over CLAHE on the adverse-weather mean.
The method is cheap: 1.5 ms CPU and 0.19 ms GPU preprocessing time, below ResNet18's own inference time, so it adds little overhead.
Because the preprocessing is independent of the classifier task and architecture, it can be combined with other training-based domain adaptation methods.
The learned target distribution is non-uniform, suggesting that the classifier prefers a specific intensity rebalancing rather than equalization.

Reading between the lines

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

A natural extension is to condition the learned target distribution on a cheap domain estimate, such as a brightness or fog detector, preserving the rank-preserving preprocessing while adapting to different camera response functions or weather types.
Because the transform discards spatial layout entirely, it may be best suited for object-level classification rather than segmentation or detection, where spatial structure matters; this is a consequence of the method's design, not a claim the paper makes.
One could train the target distribution jointly with data augmentation that simulates weather, which the paper does not do; that would test whether the learned distribution is genuinely weather-agnostic or overfit to CODaN daytime statistics.
The method's success on a specific architecture (ResNet18) may not transfer to models that rely more on absolute intensities or local texture; a cross-architecture evaluation would clarify how broadly the rank-preservation assumption holds.

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.

Desk Editor's Note

read the letter

Two things to know about this paper. First, the core idea is genuinely new and simple: make the target distribution of histogram matching a trainable vector, optimize it on normal weather images with a classifier loss, and then apply it at test time to any image. The fixed-uniform control approximating HE is a nice internal check, and the gains on fog, rain, sand, and snow (4–8 points over CLAHE) are meaningful for deployment. Second, the headline 52.53% adverse mean is not trustworthy as reported because the size of the trainable parameter vector s was selected by sweeping on the same adverse test sets used for the final table (Supplementary Table 3), not on a held-out validation split. The paper does not disclose this in the main text. This is a real validity problem for the magnitude, though not necessarily for the direction—even s=50176 gets 41.35%, slightly above HE's 40.84%, so the method likely still helps, just less than claimed.

The paper also overstates night performance: the proposed method gets 57.35% on night, below CIConv (60.95), HE (59.02), and CLAHE (58.52). The generality claim should be scoped to fog, rain, sand, and snow. There are no error bars, and the modified DAWN dataset is not released, so the differences could be within noise and the numbers are not independently reproducible.

What the paper does well: the formulation is clean, the control experiment is informative, and the computation time is small. The acknowledged noisy output images are a limitation but not a fatal one for classification.

The stress-test concern holds up on reading: the subjective selection of s on the test sets is exactly what a careful reviewer would flag. My suggestion is to resubmit with a proper validation split and report results across s, or at least add error bars and release the code.

This paper deserves serious peer review; the idea is useful and the experiments are directionally plausible. It needs revisions, not rejection. I would bring it to a reading group to discuss how common this test-set selection is.

Referee Report

3 major / 3 minor

Summary. The paper proposes a differentiable and parametric histogram matching (HM) preprocessing for image classification. Instead of matching an input image's histogram to a fixed uniform target (as in histogram equalization), the method sorts each color channel's pixels, replaces the sorted values with upsampled trainable parameters, and optimizes these parameters end-to-end with the downstream classifier's cross-entropy loss on normal-weather training images. The resulting target distribution is then applied at test time to adverse-weather images. Experiments on CODaN (day/night) and a modified DAWN dataset (fog/rain/sand/snow) report 52.53% adverse-mean top-1 accuracy, versus 40.84% for HE and 46.84% for CLAHE, with only modest computational overhead.

Significance. If the generalization claim holds, the method is a simple, fast, and potentially useful preprocessing that improves classifier robustness to several adverse weather conditions without requiring adverse-weather training data. The paper has concrete strengths: the fixed-uniform control in Table 1 (Proposed Fixed, 42.98% adverse mean) approximates HE (40.84%) and therefore supports the claim that the gain comes from the learned distribution rather than from the differentiable/parametric machinery; the complexity analysis is clear; and the supplementary explanation of why conventional HM is non-differentiable is well written. The significance is currently tempered by a load-bearing test-set selection issue and by overbroad claims about outperforming conventional preprocessing, as detailed below.

major comments (3)

[Section 4 and Supplementary B, Table 3] The choice s=2048 is not made by a validation-based protocol. Supplementary Table 3 sweeps s over {256, 512, 1024, 2048, 4096, 50176} and reports adverse-mean accuracy on exactly the CODaN night and DAWN fog/rain/sand/snow test sets used in Table 1; s=2048 is the best row. The main text simply states 'The size s of trainable parameters pc was set to 2048' with no indication that this value was selected on a held-out validation split. Consequently, the headline 52.53% adverse mean partly reflects test-set information, and the claim that the method is trained 'on only normal weather images' is incomplete because hyperparameter selection used the adverse test conditions the method claims to generalize to. This is a load-bearing validity issue for the reported magnitude. The direction may survive under a less favorable choice (s=50176 gives 41.35%, s=256 gives 40.10%), so this is not a refutation, but the paper should either select s on a proper validation split (e.g., the CODaN daytime validation set) or report the full sweep with an explicit selection rule and temper the claims accordingly.
[Table 1] The abstract and introduction state that the proposed method 'outperforms conventional preprocessing methods under adverse weather conditions,' but Table 1 shows this is not true for night images: the proposed method achieves 57.35% on night, while HE achieves 59.02%, CLAHE achieves 58.52%, and CIConv achieves 60.95%. The proposed method also reduces day accuracy relative to the no-preprocessing baseline (77.56 vs. 79.80). Thus the claim is too broad as stated; the paper should specify that improvements occur mainly on fog, rain, sand, and snow, and should discuss why night and day performance regress. This qualification matters because the adverse-mean headline is driven by large gains on sand (68.55 vs. 49.78 for HE) and does not reflect uniform improvement across adverse conditions.
[Section 3] The core design choice is to sort pixels by value, discard absolute intensities and spatial layout, and replace the sorted values with learned target quantiles. The paper provides no ablation or theoretical argument that this rank-preserving operation retains the class-discriminative information that ResNet18 needs. The only evidence is the accuracy table, which is empirical but does not isolate whether the benefit comes from the rank-preserving representation or from some other aspect of the pipeline (e.g., the specific upsampling or the per-channel parameterization). A dedicated analysis—for instance, comparing against a learnable monotonic mapping that preserves absolute intensity values, or a spatially aware variant—would strengthen the central claim that the learned distribution, rather than the sorting operation, is responsible for the generalization.

minor comments (3)

[Supplementary B, Table 3 caption] The caption contains a typo: 'prameterss' should be 'parameters', and 's = 50176corresponds' is missing a space.
[Abstract and Section 1] The phrase 'a single, well-designed distribution' is slightly misleading because the method optimizes three per-channel parameter vectors p_1, p_2, p_3, one per color channel; consider clarifying that the target is a per-channel distribution.
[Table 1] No error bars, confidence intervals, or multiple-seed results are reported. Given that the proposed method's advantage over HE/CLAHE is modest on some conditions (e.g., rain: 47.22 vs. 42.06 for CLAHE; snow: 45.27 vs. 38.49 for CLAHE), reporting standard deviations across several random initializations would help assess robustness of the conclusion.

Circularity Check

0 steps flagged · score 1.0 of 10

No circular reduction in the central generalization claim; the only caveat is a possible test-set-selected hyperparameter, which is a validity concern rather than circularity.

full rationale

The central claim is empirical: the target pixel-value distribution parameters p_c are optimized with the classifier cross-entropy loss on the CODaN daytime training set, and then the same fixed p_c is applied to held-out night and DAWN weather sets. The target distribution is not defined in terms of the adverse-weather test accuracy, so the improved adverse-mean accuracy is not a tautology. The supplementary argument that conventional HM is non-differentiable (Sec. A) is a self-contained mathematical analysis of CDF and LUT gradient behavior, not a borrowed conclusion. Self-citations [12,13] are confined to a related-work list and do not carry the argument. The only concern is that the main text fixes s=2048 without describing a validation-based selection, while Supplementary Table 3 sweeps s and reports adverse-mean accuracy on the same CODaN night and DAWN test sets used in Table 1, with s=2048 being the best row. If that sweep informed the choice, the headline 52.53% partly reflects test-set information; however, the paper does not state such a selection, and the learned p_c are independent of adverse data. This is a statistical-validity caveat, not a circular reduction.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The method introduces one learned object (the target distribution parameters) and depends on three domain assumptions about rank preservation, single-distribution transfer, and validity of the modified DAWN benchmark. No new physical or conceptual entities are posited.

free parameters (2)

target distribution parameters p_c (per color channel) = s=2048, trained via cross-entropy; initial values linearly spaced from 0 to 1
These are the learned target quantiles; the central mechanism. The paper optimizes them on CODaN daytime training data (Section 3, Section 4).
parameter count s (number of trainable histogram parameters) = 2048
Selected by the authors after an ablation on the adverse-weather test sets (Supplementary Table 3); peak at 2048, with 4096 and 50176 worse. This is a test-set-driven hyperparameter.

assumptions (4)

domain assumption Sorting pixels by value and replacing them with learned target quantiles preserves the information the classifier needs for classification (Section 3, after the definition of x'_c).
The entire method discards absolute pixel intensities and spatial layout. The paper provides no proof or dedicated ablation that rank-order statistics alone retain class-discriminative cues; the accuracy table is the only evidence.
domain assumption A single fixed target distribution can normalize the diverse histograms of unseen adverse-weather domains (Section 1 introduction, Section 4 evaluation).
The method assumes distribution shape is the dominant nuisance for cross-weather generalization; the paper tests this on CODaN night and four DAWN weathers, not on a broad set of domains.
domain assumption The modified DAWN classification set, built from detection crops with zero padding and manual exclusion of ambiguous or overlapped objects, is a valid and comparable benchmark to CODaN classification (Section 4, Fig. 3).
Accuracy numbers depend on the crop and exclusion pipeline; the dataset is not released, so the assumption is not externally checkable.
standard math Standard deep-learning training assumptions: SGD, cross-entropy loss, learned classifier parameters (Section 4).
No unusual mathematical axioms; the backpropagation framework is standard.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Rethinking Image Histogram Matching for Image Classification." pith.science (2026). https://pith.science/paper/LDKN456M

@misc{pith2026250601346,
  author       = {Pith},
  title        = {Pith review of: Rethinking Image Histogram Matching for Image Classification},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LDKN456M}},
  note         = {Machine review of arXiv:2506.01346}
}

read the original abstract

This paper rethinks image histogram matching (HM) and proposes a differentiable and parametric HM preprocessing for a downstream classifier. Convolutional neural networks have demonstrated remarkable achievements in classification tasks. However, they often exhibit degraded performance on low-contrast images captured under adverse weather conditions. To maintain classifier performance under low-contrast images, histogram equalization (HE) is commonly used. HE is a special case of HM using a uniform distribution as a target pixel value distribution. In this paper, we focus on the shape of the target pixel value distribution. Compared to a uniform distribution, a single, well-designed distribution could have potential to improve the performance of the downstream classifier across various adverse weather conditions. Based on this hypothesis, we propose a differentiable and parametric HM that optimizes the target distribution using the loss function of the downstream classifier. This method addresses pixel value imbalances by transforming input images with arbitrary distributions into a target distribution optimized for the classifier. Our HM is trained on only normal weather images using the classifier. Experimental results show that a classifier trained with our proposed HM outperforms conventional preprocessing methods under adverse weather conditions.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

31 extracted references · 30 canonical work pages

[1]

Rethinking Image Histogram Matching for Image Classification

INTRODUCTION Convolutional neural networks (CNNs) have achieved re- markable performance in many computer vision applications [1, 2]. Most CNNs are trained on images under ideal con- ditions, such as daytime and clear weather. However, the performance of CNNs often degrades when processing low- contrast images captured under adverse weather conditions, su...

work page Pith review arXiv
[2]

LUT and curve-based methods are particularly studied for image- adaptive methods

RELA TED WORK Tone mapping methods based on LUT, curve, and histogram are generally used to enhance the contrast of images [5]. LUT and curve-based methods are particularly studied for image- adaptive methods. These image-adaptive methods use neural networks to estimate tone mapping parameters for each input image, which are then used to enhance image con...

work page
[3]

1 shows a block diagram of the proposed HM

PROPOSED METHOD Fig. 1 shows a block diagram of the proposed HM. We pre- pare trainable parameters following the target distribution and use them to replace all pixel values in the input image. We sort the pixels using their pixel values as keys and maintain their relative ordering during this replacement. Fig. 2 illus- trates the difference between the c...

work page
[4]

The CODaN dataset is designed for classification and contains 10 classes of day- time and nighttime images

EXPERIMENTS To evaluate the performance of the proposed method in ad- verse weather conditions, we used two datasets: CODaN [2] and a modified version of DAWN [21]. The CODaN dataset is designed for classification and contains 10 classes of day- time and nighttime images. This dataset consists of cropped images from object detection datasets. CODaN includ...

work page
[5]

We redesigned the HM to be differentiable and parametric

CONCLUSION This paper proposed an optimization method for image pixel value distributions to enhance the performance of down- stream classifiers. We redesigned the HM to be differentiable and parametric. The proposed method can optimize the target distribution using a loss function of the downstream classifier. Experimental results confirmed that the prop...

work page
[6]

Deep residual learning for image recognition,

Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun, “Deep residual learning for image recognition,” in IEEE Conf. Comput. Vis. Pattern Recog. , 2016, pp. 770–778

work page 2016
[7]

Zero-shot day-night domain adap- tation with a physics prior,

Attila Lengyel, Sourav Garg, Michael Milford, and Jan C. van Gemert, “Zero-shot day-night domain adap- tation with a physics prior,” in Int. Conf. Comput. Vis. , 2021, pp. 4379–4389

work page 2021
[8]

Zero-reference deep curve estimation for low-light im- age enhancement,

Chunle Guo, Chongyi Li, Jichang Guo, Chen Change Loy, Junhui Hou, Sam Kwong, and Runmin Cong, “Zero-reference deep curve estimation for low-light im- age enhancement,” in IEEE Conf. Comput. Vis. Pattern Recog., 2020, pp. 1777–1786

work page 2020

Show all 31 references

[9]

Watching it in dark: A target-aware representation learning frame- work for high-level vision tasks in low illumination,

Yunan Li, Yihao Zhang, Shoude Li, Long Tian, Dou Quan, Chaoneng Li, and Qiguang Miao, “Watching it in dark: A target-aware representation learning frame- work for high-level vision tasks in low illumination,” in Eur . Conf. Comput. Vis., 2024, pp. 37–53

2024
[10]

Intensity transformations and spa- tial filtering,

Rafael C. Gonzalez, “Intensity transformations and spa- tial filtering,” in Digital Image Processing , chapter 3, pp. 119–153. Pearson Deutschland, 4th edition, 2022

2022
[11]

Learning to enhance low-light image via zero- reference deep curve estimation,

Chongyi Li, Chunle Guo, and Chen Change Loy, “Learning to enhance low-light image via zero- reference deep curve estimation,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 44, no. 8, pp. 4225–4238, 2022

2022
[12]

Learning image-adaptive 3D lookup tables for high performance photo enhancement in real-time,

Hui Zeng, Jianrui Cai, Lida Li, Zisheng Cao, and Lei Zhang, “Learning image-adaptive 3D lookup tables for high performance photo enhancement in real-time,” IEEE Trans. Pattern Anal. Mach. Intell. , vol. 44, no. 4, pp. 2058–2073, 2022

2022
[13]

Toward raw object detection: A new bench- mark and a new model,

Ruikang Xu, Chang Chen, Jingyang Peng, Cheng Li, Yibin Huang, Fenglong Song, Youliang Yan, and Zhiwei Xiong, “Toward raw object detection: A new bench- mark and a new model,” in IEEE Conf. Comput. Vis. Pattern Recog., 2023, pp. 13384–13393

2023
[14]

Self-aligned concave curve: Illumination enhancement for unsupervised adaptation,

Wenjing Wang, Zhengbo Xu, Haofeng Huang, and Ji- aying Liu, “Self-aligned concave curve: Illumination enhancement for unsupervised adaptation,” in ACM Int. Conf. Multimedia, 2022, pp. 2617–2626

2022
[15]

Unsupervised illumination adaptation for low- light vision,

Wenjing Wang, Rundong Luo, Wenhan Yang, and Jiay- ing Liu, “Unsupervised illumination adaptation for low- light vision,” IEEE Trans. Pattern Anal. Mach. Intell. , vol. 46, no. 9, pp. 5951–5966, 2024

2024
[16]

Deep degradation prior for low-quality image classification,

Yang Wang, Yang Cao, Zheng-Jun Zha, Jing Zhang, and Zhiwei Xiong, “Deep degradation prior for low-quality image classification,” in IEEE Conf. Comput. Vis. Pat- tern Recog., 2020, pp. 11046–11055

2020
[17]

ERUP-YOLO: Enhancing object detection robust- ness for adverse weather condition by unified image- adaptive processing,

Yuka Ogino, Yuho Shoji, Takahiro Toizumi, and Atsushi Ito, “ERUP-YOLO: Enhancing object detection robust- ness for adverse weather condition by unified image- adaptive processing,” arXiv preprint arXiv:2411.02799, 2024

2024 arXiv
[18]

Improving low-light image recognition performance based on image-adaptive learn- able module,

Seitaro Ono, Yuka Ogino, Takahiro Toizumi, Atsushi Ito, and Masato Tsukada, “Improving low-light image recognition performance based on image-adaptive learn- able module,” arXiv preprint arXiv:2401.06438, 2025

2025 arXiv
[19]

Raw or cooked? object detection on raw images,

William Ljungbergh, Joakim Johnander, Christoffer Pe- tersson, and Michael Felsberg, “Raw or cooked? object detection on raw images,” in Scandinavian Conf. Image Anal., 2023, pp. 374–385

2023
[20]

Similarity min-max: Zero-shot day-night do- main adaptation,

Rundong Luo, Wenjing Wang, Wenhan Yang, and Jiay- ing Liu, “Similarity min-max: Zero-shot day-night do- main adaptation,” in Int. Conf. Comput. Vis. , 2023, pp. 8070–8080

2023
[21]

Contrast limited adaptive histogram equalization,

Karel Zuiderveld, “Contrast limited adaptive histogram equalization,” in Graphics Gems IV, pp. 474–485. Aca- demic Press Professional, Inc., USA, 1994

1994
[22]

Face recognition in low lighting conditions using fisherface method and CLAHE techniques,

Muhammad Fauzan Rahman, Febryanti Sthevanie, and Kurniawan Nur Ramadhani, “Face recognition in low lighting conditions using fisherface method and CLAHE techniques,” in Int. Conf. Inf. Commun. Technol. , 2020, pp. 1–6

2020
[23]

Histogram learning in image contrast enhance- ment,

Bin Xiao, Yunqiu Xu, Han Tang, Xiuli Bi, and Weisheng Li, “Histogram learning in image contrast enhance- ment,” in IEEE Conf. Comput. Vis. Pattern Recog. Worksh., 2019, pp. 1880–1889

2019
[24]

Restoring images in adverse weather conditions via histogram transformer,

Shangquan Sun, Wenqi Ren, Xinwei Gao, Rui Wang, and Xiaochun Cao, “Restoring images in adverse weather conditions via histogram transformer,” in Eur . Conf. Comput. Vis., 2024, pp. 111–129

2024
[25]

Histoformer: Histogram-based transformer for efficient underwater image enhance- ment,

Yan-Tsung Peng, Yen-Rong Chen, Guan-Rong Chen, and Chun-Jung Liao, “Histoformer: Histogram-based transformer for efficient underwater image enhance- ment,” IEEE J. Ocean. Eng., vol. 50, no. 1, pp. 164–177, 2025

2025
[26]

DAWN: vehicle detection in adverse weather nature dataset,

Mourad A. Kenk and Mahmoud Hassaballah, “DAWN: vehicle detection in adverse weather nature dataset,” arXiv preprint arXiv:2008.05402, 2020

2008 arXiv
[27]

Au- tomatic differentiation in pytorch,

Adam Paszke, Sam Gross, Soumith Chintala, Gregory Chanan, Edward Yang, Zachary DeVito, Zeming Lin, Alban Desmaison, Luca Antiga, and Adam Lerer, “Au- tomatic differentiation in pytorch,” in Adv. Neural In- form. Process. Syst. Worksh., 2017

2017
[28]

Learn- ing deep embeddings with histogram loss,

Evgeniya Ustinova and Victor Lempitsky, “Learn- ing deep embeddings with histogram loss,” in Adv. Neural Inform. Process. Syst. , D. Lee, M. Sugiyama, U. Luxburg, I. Guyon, and R. Garnett, Eds., 2016, vol. 29

2016
[29]

Learnable histogram: Statistical context fea- tures for deep neural networks,

Zhe Wang, Hongsheng Li, Wanli Ouyang, and Xiaogang Wang, “Learnable histogram: Statistical context fea- tures for deep neural networks,” in Eur . Conf. Comput. Vis., Cham, 2016, pp. 246–262

2016
[30]

Differentiable histogram with hard-binning,

Ibrahim Yusuf, George Igwegbe, and Oluwafemi Azeez, “Differentiable histogram with hard-binning,” arXiv preprint arXiv:2012.06311, 2020

2012 arXiv
[31]

Differentiable histogram loss functions for intensity-based image-to-image translation,

Mor Avi-Aharon, Assaf Arbelle, and Tammy Rik- lin Raviv, “Differentiable histogram loss functions for intensity-based image-to-image translation,” IEEE Trans. Pattern Anal. Mach. Intell. , vol. 45, no. 10, pp. 11642–11653, 2023

2023

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [1]

Rethinking Image Histogram Matching for Image Classification

INTRODUCTION Convolutional neural networks (CNNs) have achieved re- markable performance in many computer vision applications [1, 2]. Most CNNs are trained on images under ideal con- ditions, such as daytime and clear weather. However, the performance of CNNs often degrades when processing low- contrast images captured under adverse weather conditions, su...

work page Pith review arXiv

[2] [2]

LUT and curve-based methods are particularly studied for image- adaptive methods

RELA TED WORK Tone mapping methods based on LUT, curve, and histogram are generally used to enhance the contrast of images [5]. LUT and curve-based methods are particularly studied for image- adaptive methods. These image-adaptive methods use neural networks to estimate tone mapping parameters for each input image, which are then used to enhance image con...

work page

[3] [3]

1 shows a block diagram of the proposed HM

PROPOSED METHOD Fig. 1 shows a block diagram of the proposed HM. We pre- pare trainable parameters following the target distribution and use them to replace all pixel values in the input image. We sort the pixels using their pixel values as keys and maintain their relative ordering during this replacement. Fig. 2 illus- trates the difference between the c...

work page

[4] [4]

The CODaN dataset is designed for classification and contains 10 classes of day- time and nighttime images

EXPERIMENTS To evaluate the performance of the proposed method in ad- verse weather conditions, we used two datasets: CODaN [2] and a modified version of DAWN [21]. The CODaN dataset is designed for classification and contains 10 classes of day- time and nighttime images. This dataset consists of cropped images from object detection datasets. CODaN includ...

work page

[5] [5]

We redesigned the HM to be differentiable and parametric

CONCLUSION This paper proposed an optimization method for image pixel value distributions to enhance the performance of down- stream classifiers. We redesigned the HM to be differentiable and parametric. The proposed method can optimize the target distribution using a loss function of the downstream classifier. Experimental results confirmed that the prop...

work page

[6] [6]

Deep residual learning for image recognition,

Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun, “Deep residual learning for image recognition,” in IEEE Conf. Comput. Vis. Pattern Recog. , 2016, pp. 770–778

work page 2016

[7] [7]

Zero-shot day-night domain adap- tation with a physics prior,

Attila Lengyel, Sourav Garg, Michael Milford, and Jan C. van Gemert, “Zero-shot day-night domain adap- tation with a physics prior,” in Int. Conf. Comput. Vis. , 2021, pp. 4379–4389

work page 2021

[8] [8]

Zero-reference deep curve estimation for low-light im- age enhancement,

Chunle Guo, Chongyi Li, Jichang Guo, Chen Change Loy, Junhui Hou, Sam Kwong, and Runmin Cong, “Zero-reference deep curve estimation for low-light im- age enhancement,” in IEEE Conf. Comput. Vis. Pattern Recog., 2020, pp. 1777–1786

work page 2020