Pith. sign in

REVIEW 4 major objections 4 minor 29 references

CURVE: CLIP-Utilized Reinforcement Learning for Visual Image Enhancement via Simple Image Processing

T0 review · 4 major / 4 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 CURVE claims that zero-reference image enhancement can be reduced to a sequence of global tone-curve adjustments, chosen by a reinforcement-learning policy and rewarded by CLIP text similarity, and that this simple recipe matches or beats…

desk verdict A pragmatic RL+CLIP tone-curve method with a clever LUT speedup; the evaluation is mostly solid but the RL ablation is unfair and the SICE text contradicts its own table. read the letter →

arxiv 2505.23102 v2 pith:S5WAKHOB submitted 2025-05-29 cs.CV

Yuka Ogino , Takahiro Toizumi , Atsushi Ito This is my paper
classification cs.CV
keywords low-lightimageenhancementzero-referencereinforcementlearningCLIPBeziercurvetonemappinglook-uptableSoftActor-Critic
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

CURVE claims that zero-reference low-light and multi-exposure image enhancement can be reduced to a sequence of global tone-curve adjustments chosen by a reinforcement-learning policy. The reward is not a hand-designed image statistic but the agreement, in CLIP embedding space, between the enhanced image and text prompts such as "a good photo of {object class}", contrasted with a fixed negative prompt. Because the policy operates on a small 56×56 state and the final tone mapping is applied through a composed lookup table, the method reports roughly constant processing time across HD, FHD, and UHD resolutions, about 0.017 seconds per frame. On LOLv1, LOLv2Real, and LOLv2Syn the method reaches the best or second-best SSIM and PSNR among the zero-reference baselines compared, and on the multi-exposure SICE dataset it keeps brightness consistent across exposure levels. A sympathetic reading is that perceptual language supervision plus iterative simple processing is a viable alternative to heavier learned networks.

What carries the argument

The load-bearing object is the Bezier-curve tone adjustment module: a cubic Bezier curve with endpoints fixed at $(0,0)$ and $(1,1)$, whose two interior control points are moved by four action parameters, and which is evaluated as a piecewise-linear mapping so a full-resolution image can be transformed by a single lookup-table application. Around this module, the paper builds a Soft Actor-Critic (SAC) loop, an off-policy maximum-entropy reinforcement learning algorithm, in which the state is the concatenation of the current and previous small images ($x_t$ and $x_t - x_{t-1}$), the action is the four-parameter curve adjustment, and the reward is the decrease in a CLIP-based softmax cross-entropy loss that pushes the image embedding toward "a good photo of {class}" and away from "a bad, saturated, blacked out photo of nothing". The test-time trick is to apply the learned sequence of actions to a lookup table over all $2^{bit}$ pixel values rather than to the image itself, so the whole episode composes into one LUT that maps the original high-resolution image directly to the final result.

What would settle it

Train CURVE exactly as described but replace the CLIP reward with a reward that only maximizes global brightness or contrast; if the resulting images score the same on LOLv1 SSIM and PSNR, then the CLIP text supervision is not essential. Conversely, run the trained policy on real low-light images with heavy sensor noise: since the action is a global tone curve, any large drop in perceived quality relative to clean synthetic test images would show that the synthetic Bezier-augmentation training distribution, not the CLIP reward, is the brittle link.

Watch

Extended reading notes

Core claim

On its own terms, the paper's claim is that CURVE establishes a new point in the design space of zero-reference low-light image enhancement: use a frozen CLIP model to define what "good" means, train a Soft Actor-Critic policy to take repeated small actions, and let each action modify the control points of a global cubic Bezier tone curve. The action vector $a_t = [\theta_1, \theta_2, r_1, r_2]$ is parameterized so that zero actions leave the image unchanged, and the reward $r_t = \beta(L_t - L_{t+1})$ rewards any step that reduces a softmax cross-entropy loss between the CLIP image embedding and positive and negative text embeddings. The policy is trained on VOC 2007 images with randomly sampled Bezier augmentations and tested directly on low-light and multi-exposure datasets. The reported result is either the best or second-best SSIM and PSNR among the compared zero-reference methods on all three LOL datasets, with a processing speed an order of magnitude faster than most baselines at high resolutions because the state is tiny and the full-resolution image is never processed until the final lookup-table application.

Load-bearing premise

The load-bearing premise is that a policy trained on VOC 2007 images randomly darkened or brightened with Bezier curves drawn from N(0,1), and rewarded by CLIP text similarity, will enhance real low-light and multi-exposure images without any real paired or unpaired low-light training data.

Editorial extensions

If this is right

  • Because the policy and Q-networks process only a 56×56 state, test-time cost is nearly independent of resolution; the paper reports 0.017 s/frame for HD, FHD, and UHD on an RTX 3080, versus 0.205 s for Zero-DCE and 20.68 s for CLIP-LIT at UHD.
  • The lookup-table composition means arbitrarily many iterative adjustments can be applied to a high-resolution image in one pass, so the speed advantage should persist for longer episodes.
  • Training on VOC 2007 with random Bezier-curve augmentations transfers to LOLv1, LOLv2Real, LOLv2Syn, and SICE, suggesting that synthetic tone-curve degradations can stand in for real low-light training data.
  • The RL formulation keeps improving with more steps and stays stable, while the train-by-loss ablation peaks and then degrades (Fig. 4), so the reward-as-improvement design is doing real work.
  • The same framework handles under-exposed and over-exposed images on SICE, because the policy learns to move brightness in the direction that lowers the CLIP loss.

Reading between the lines

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

  • If CLIP-text similarity is the right perceptual proxy, the same reward could steer other global image operators such as white balance, saturation, or gamma by swapping only the differentiable processing module, since the RL loop is operator-agnostic.
  • The reported UHD speed of 0.017 s/frame implies roughly 60 frames per second, so a frame-wise video extension is plausible; the paper does not address temporal flicker, which would be the next obstacle.
  • The positive prompt is built from object classes detected in the training image, coupling enhancement to recognition; this may make CURVE especially suited to preprocessing for detection, at the possible cost of human-perceptual quality on scenes with no clear object.
  • The global tone curve cannot repair local contrast or remove noise; the paper itself notes amplification of noise and tonal differences on over-exposed images, so a spatially varying extension would need to sacrifice the LUT speed trick.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 4 minor

Summary. The paper proposes CURVE, a zero-reference low-light image enhancement method that combines Soft Actor-Critic (SAC) reinforcement learning with a global Bezier tone-curve adjustment module. The reward is computed from CLIP text-image similarity, using object-class-conditioned positive prompts and a fixed negative prompt. At test time, the policy is run on a small downsampled image, and a composed look-up table (LUT) is applied to the full-resolution image. The paper reports SSIM and PSNR results on LOLv1, LOLv2Real, LOLv2Syn, and SICE Part 2, claiming best or second-best performance among compared zero-reference methods while being substantially faster on HD/FHD/UHD resolutions. The main claim is that a simple iterative global tone-curve policy trained with a CLIP-based reward can match or outperform conventional CNN-based enhancers without paired data.

Significance. If the reported results hold, the paper makes a practical contribution: it demonstrates that a compact global tone-curve policy can compete with much larger CNN-based zero-reference enhancers on standard low-light benchmarks, and the LUT-based implementation gives a genuine speed advantage on high-resolution images. The use of CLIP as a reward signal with object-conditioned prompts is a sensible and reasonably novel design, and the SAC formulation is clearly described. The paper does not provide code or trained models, but the method is simple enough to re-implement. The main risk is that the policy is trained exclusively on VOC 2007 images augmented with Bezier curves sampled from N(0,1), and the paper does not establish that the CLIP reward on these synthetic augmentations tracks perceptual quality on real under- or over-exposed photographs. This transfer assumption is load-bearing for the claimed zero-reference generalization.

major comments (4)
  1. [Sec. 3.1 and Sec. 3.2] The policy is trained only on VOC 2007 images augmented with Bezier tone curves sampled from N(0,1), yet it is evaluated on real low-light (LOLv1/v2) and multi-exposure (SICE) images without any domain adaptation or analysis. The paper does not show that maximizing the CLIP reward on this synthetic augmentation family tracks perceptual quality on real illumination defects. I ask for (i) an ablation trained on real unlabeled low-light images, (ii) a sensitivity analysis with different augmentation distributions, and (iii) a held-out synthetic validation demonstrating that the learned reward correlates with PSNR/SSIM. Without this, the reported zero-reference gains could be a distributional artifact of the specific N(0,1) curve family rather than a general enhancement capability.
  2. [Sec. 3.2, Table 4] The sentence 'Our approach is outperformed by the baseline methods' directly contradicts Table 4, in which CURVE is better than Zero-DCE on SSIM, PSNR, and LPIPS. Please correct the text and state which comparison is intended. As written, the paper undercuts its own SICE result and leaves the reader uncertain about which numbers are trustworthy.
  3. [Sec. 3.1] The comparison with 'train-by-loss' is not equal in training effort: CURVE is trained for 7.5e5 SAC iterations, while train-by-loss uses only 15k iterations. Moreover, no standard deviations or multiple-seed results are reported for any method. Since SAC is stochastic and the reported margins are small (e.g., SSIM 0.7164 versus 0.7035 on LOLv1), a single run does not establish superiority. Please report mean and standard deviation over at least three seeds and align the iteration count for the ablation.
  4. [Sec. 2.2.2] The reward uses object-class-conditioned positive prompts, but the paper does not specify how object classes are detected at test time. Is a pretrained detector used, or are VOC ground-truth labels assumed? For LOL and SICE images, no such labels exist, so this detail is essential for reproducing the method. The paper also does not state what happens when N=0 (no detected classes), in which case Eq. (6) is undefined.
minor comments (4)
  1. [Sec. 2.2.2 and Fig. 7] The negative prompt text is written as 'a bad, saturated, blacked out photo of nothing' in the main text but 'a bad, saturated and blacked-out photo of nothing' in Fig. 7(b); please make the wording consistent.
  2. [Sec. 2.2.3 and Table 2] The policy samples actions from a Gaussian and then applies tanh and rescaling, but the rescaling factor that maps tanh outputs to the claimed action range [-2, 2] is not stated. Please clarify.
  3. [Sec. 2.3, Algorithm 1] The operation B(l_t, a_t) on a LUT vector is not defined; the Bezier module B was introduced for images, so please explain how it is applied to a 1-D lookup table.
  4. [Table 1 and Sec. 3.1] The runtime for ReLLIE is measured on CPU while all other methods are measured on GPU, making the speed comparison for ReLLIE not directly comparable. Please add an explicit caveat in the table or text.

Circularity Check

0 steps flagged · score 1.0 of 10

No load-bearing circularity: the CLIP reward is an external fixed model and the reported SSIM/PSNR numbers are external to that reward, so the central claim does not reduce to its inputs.

full rationale

CURVE's derivation chain is not circular. The policy is trained against a reward built from CLIP text-image similarity (Eqs. 6-7), where CLIP is a fixed pretrained external model and the prompts are generic ("a good photo of {class}" vs "a bad, saturated, blacked out photo of nothing"). The reported claims (Tables 1 and 4) are evaluated with SSIM, PSNR, and LPIPS against ground-truth references, which are not the training reward and are not fitted parameters. The Bézier tone-curve module is adapted from the authors' prior work [11], a self-citation, but it is an independently published parameterization, is not fitted to the target datasets, and is not invoked as a uniqueness theorem. The paper also reports a "train-by-loss" ablation that isolates the RL contribution, showing that the central comparison is between two training objectives rather than a definitional identity. There is an internal textual inconsistency in Sec. 3.2 where "Our approach is outperformed by the baseline methods" contradicts Table 4, but that is a correctness/consistency issue, not circularity. The assumption that an N(0,1) curve augmentation on VOC images transfers to real low-light and multi-exposure data is a distributional assumption, not a circular reduction. No equation defines the measured metrics in terms of the CLIP reward, and no fitted parameter is later renamed a prediction. The paper's self-stated limitation, that global processing "may amplify noise or tonal differences," is also consistent with a non-circular design. Only a minor, non-load-bearing self-citation is present, so the circularity score is 1.

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

The method relies on CLIP as a perceptual oracle, on synthetic Bezier-augmented VOC images as a training distribution, and on a heavily downsampled state; none of these are validated independently inside the paper, and the RL-vs-loss ablation is confounded by a 50x iteration gap.

free parameters (4)
  • beta (reward scaling factor) = 200
    Scaling factor in Eq. 7, chosen by hand; directly sets the magnitude of rewards the SAC agent sees.
  • T (episode steps) = 5
    Number of tone curve iterations per episode; set to 5 in Table 2 (and 20 for the Fig. 4 analysis); changes the expressiveness of the final composed curve.
  • Action value range = [-2, 2]
    tanh output is rescaled to this range (Table 2), limiting the Bezier control point positions; chosen by hand.
  • L (Bezier segments per curve) = unspecified
    Sec. 2.1 partitions parametric space into L+1 points but L is not reported; the piecewise-linear approximation error depends on it.
assumptions (4)
  • domain assumption CLIP text-image similarity is a valid reward proxy for perceptual enhancement quality.
    Eq. 6 and Eq. 7 define reward entirely through CLIP cosine similarities between the enhanced image and prompts like 'a good photo of cat'; no human preference data grounds this assumption beyond CLIP's pretraining.
  • domain assumption Training on VOC 2007 with random Bezier-curve tone augmentation transfers to real low-light and multi-exposure images.
    Sec. 3.1 trains on VOC 2007 trainval augmented with synthetic Bezier curves drawn from N(0,1); no real low-light images are used in training, so transfer is assumed.
  • domain assumption The 56x56 downsampled state (concatenated current and previous frames) carries enough information to predict good global tone curve parameters.
    Sec. 2.2.1 defines s_t in R^{2C x h x w} with h=w=56; the policy never sees the full resolution image during training, so local detail relevant to exposure is discarded.
  • standard math Sequential global tone curve applications compose into a single LUT that faithfully reproduces the per-step processing on the original image.
    Algorithm 1 applies the same per-pixel mapping to a LUT; this is exact because Eq. 3 is a global per-pixel intensity mapping, assuming no spatial or chromatic cross-talk.

how reviews work

0 comments
Cite this review

Pith. "Pith review of CURVE: CLIP-Utilized Reinforcement Learning for Visual Image Enhancement via Simple Image Processing." pith.science (2026). https://pith.science/paper/S5WAKHOB

@misc{pith2026250523102,
  author       = {Pith},
  title        = {Pith review of: CURVE: CLIP-Utilized Reinforcement Learning for Visual Image Enhancement via Simple Image Processing},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/S5WAKHOB}},
  note         = {Machine review of arXiv:2505.23102}
}
read the original abstract

Low-Light Image Enhancement (LLIE) is crucial for improving both human perception and computer vision tasks. This paper addresses two challenges in zero-reference LLIE: obtaining perceptually 'good' images using the Contrastive Language-Image Pre-Training (CLIP) model and maintaining computational efficiency for high-resolution images. We propose CLIP-Utilized Reinforcement learning-based Visual image Enhancement (CURVE). CURVE employs a simple image processing module which adjusts global image tone based on B\'ezier curve and estimates its processing parameters iteratively. The estimator is trained by reinforcement learning with rewards designed using CLIP text embeddings. Experiments on low-light and multi-exposure datasets demonstrate the performance of CURVE in terms of enhancement quality and processing speed compared to conventional methods.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

29 extracted references · 26 canonical work pages

  1. [1]

    This degradation impacts both human per- ception and performance of computer vision tasks such as image recognition and object detection

    INTRODUCTION Lighting conditions, especially low-light scenarios, degrade image contrast. This degradation impacts both human per- ception and performance of computer vision tasks such as image recognition and object detection. To address these problems, Low-Light Image Enhancement (LLIE) methods have been proposed. Recent research has focused on zero- re...

    work page

  2. [2]

    PROPOSED METHOD Our proposed CURVE employs a reinforcement learning (RL) framework to iteratively apply tone curve adjustments based on B ´ezier curves. Fig. 1(a) illustrates an overview of our approach. CURVE processes an input image throughT steps of tone curve adjustments within an episode. At each time stept, our framework obtains the current states t...

    work page 1920

  3. [3]

    Experimental Settings Implementations and training details.We used the VOC 2007 trainval dataset [22] for training

    EXPERIMENTS 3.1. Experimental Settings Implementations and training details.We used the VOC 2007 trainval dataset [22] for training. Following the training method in prior work [11], we employed the tone curve mod- ule for data augmentation. We applied random parameters by sampling from a normal distributionN(0,1). The hid- den layers of MLPs and the outp...

    work page 2007

  4. [4]

    Our approach addresses both low-light and multi-exposure image enhancement challenges while maintaining computational efficiency

    CONCLUSION In this paper, we introduced CURVE, a novel image en- hancement method combining reinforcement learning with CLIP-based reward design. Our approach addresses both low-light and multi-exposure image enhancement challenges while maintaining computational efficiency. CURVE utilizes a fast, adaptive processing framework with B ´ezier curve- based t...

    work page

  5. [5]

    Iterative prompt learning for unsupervised backlit image enhancement,

    Zhexin Liang, Chongyi Li, Shangchen Zhou, Ruicheng Feng, and Chen Change Loy, “Iterative prompt learning for unsupervised backlit image enhancement,” inICCV, October 2023, pp. 8094–8103

    work page 2023

  6. [6]

    Retinex-inspired unrolling with co- operative prior architecture search for low-light image enhancement,

    Risheng Liu, Long Ma, Jiaao Zhang, Xin Fan, and Zhongxuan Luo, “Retinex-inspired unrolling with co- operative prior architecture search for low-light image enhancement,” inCVPR, June 2021, pp. 10561–10570

    work page 2021

  7. [7]

    Toward fast, flexible, and robust low- light image enhancement,

    Long Ma, Tengyu Ma, Risheng Liu, Xin Fan, and Zhongxuan Luo, “Toward fast, flexible, and robust low- light image enhancement,” inCVPR, June 2022, pp. 5637–5646

    work page 2022

  8. [8]

    CURVE: CLIP-Utilized Reinforcement Learning for Visual Image Enhancement via Simple Image Processing

    compares image embeddings with text embeddings of ”a good photo” and ”a bad photo” to derive IQA scores. It has achieved competitive performance across various IQA bench- marks. Furthermore, a CLIP-based color control method [9] has been proposed that enables color adjustments based on perceptual text expressions. However, the direct use of text prompts f...

    work page Pith review arXiv 2025

Show all 29 references
  1. [9]

    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,” inCVPR, June 2020, pp. 1777–1786

    2020

  2. [10]

    Learning transferable vi- sual models from natural language supervision,

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, and Ilya Sutskever, “Learning transferable vi- sual models from natural language supervision,”arXiv preprint ar...

    2021 arXiv

  3. [11]

    Unsupervised image prior via prompt learning and CLIP semantic guidance for low-light image en- hancement,

    Igor Morawski, Kai He, Shusil Dangi, and Winston H. Hsu, “Unsupervised image prior via prompt learning and CLIP semantic guidance for low-light image en- hancement,” inCVPRW, June 2024, pp. 5971–5981

    2024

  4. [12]

    Leveraging content and context cues for low-light image enhancement,

    Igor Morawski, Kai He, Shusil Dangi, and Winston H. Hsu, “Leveraging content and context cues for low-light image enhancement,”arXiv preprint arXiv:2412.07693, 2024

    2024 arXiv

  5. [13]

    Exploring CLIP for assessing the look and feel of im- ages,

    Jianyi Wang, Kelvin CK Chan, and Chen Change Loy, “Exploring CLIP for assessing the look and feel of im- ages,” inAAAI, 2023

    2023

  6. [14]

    CLIPtone: Unsupervised learning for text-based image tone adjustment,

    Hyeongmin Lee, Kyoungkook Kang, Jungseul Ok, and Sunghyun Cho, “CLIPtone: Unsupervised learning for text-based image tone adjustment,” inCVPR, June 2024, pp. 2942–2951

    2024

  7. [15]

    Discover- ing an image-adaptive coordinate system for photogra- phy processing,

    Ziteng Cui, Lin Gu, and Tatsuya Harada, “Discover- ing an image-adaptive coordinate system for photogra- phy processing,” inBMVC. 2024, BMV A

    2024

  8. [16]

    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

  9. [17]

    Learning to control camera exposure via reinforcement learning,

    Kyunghyun Lee, Ukcheol Shin, and Byeong-Uk Lee, “Learning to control camera exposure via reinforcement learning,” inCVPR, June 2024, pp. 2975–2983

    2024

  10. [18]

    AdaptiveISP: Learning an adaptive image signal processor for object detection,

    Yujin Wang, Tianyi Xu, Fan Zhang, Tianfan Xue, and Jinwei Gu, “AdaptiveISP: Learning an adaptive image signal processor for object detection,” inNeurIPS, 2024

    2024

  11. [19]

    Exposure: A white-box photo post- processing framework,

    Yuanming Hu, Hao He, Chenxi Xu, Baoyuan Wang, and Stephen Lin, “Exposure: A white-box photo post- processing framework,”ACM TOG, vol. 37, no. 2, pp. 26, 2018

    2018

  12. [20]

    Rellie: Deep reinforcement learning for cus- tomized low-light image enhancement,

    Rongkai Zhang, Lanqing Guo, Siyu Huang, and Bihan Wen, “Rellie: Deep reinforcement learning for cus- tomized low-light image enhancement,” inACM MM, 2021, pp. 2429–2437

    2021

  13. [21]

    Soft actor-critic: Off-policy maximum entropy deep reinforcement learning with a stochastic actor,

    Tuomas Haarnoja, Aurick Zhou, Pieter Abbeel, and Sergey Levine, “Soft actor-critic: Off-policy maximum entropy deep reinforcement learning with a stochastic actor,” inPMLR, 10–15 Jul 2018, vol. 80, pp. 1861– 1870

    2018

  14. [22]

    Soft actor-critic algorithms and applications,

    Tuomas Haarnoja, Aurick Zhou, Kristian Hartikainen, George Tucker, Sehoon Ha, Jie Tan, Vikash Kumar, Henry Zhu, Abhishek Gupta, Pieter Abbeel, et al., “Soft actor-critic algorithms and applications,”arXiv preprint arXiv:1812.05905, 2018

    2018 arXiv

  15. [23]

    Getting to know low-light images with the exclusively dark dataset,

    Yuen Peng Loh and Chee Seng Chan, “Getting to know low-light images with the exclusively dark dataset,” Computer Vision and Image Understanding, vol. 178, pp. 30–42, 2019

    2019

  16. [24]

    Deep retinex decomposition for low-light en- hancement,

    Wei Chen, Wang Wenjing, Yang Wenhan, and Liu Ji- aying, “Deep retinex decomposition for low-light en- hancement,” inBMVC. 2018, BMV A

    2018

  17. [25]

    Sparse gradient regularized deep retinex network for robust low-light image en- hancement,

    Wenhan Yang, Wenjing Wang, Haofeng Huang, Shiqi Wang, and Jiaying Liu, “Sparse gradient regularized deep retinex network for robust low-light image en- hancement,”IEEE TIP, vol. 30, pp. 2072–2086, 2021

    2021

  18. [26]

    Learn- ing a deep single image contrast enhancer from multi- exposure images,

    Jianrui Cai, Shuhang Gu, and Lei Zhang, “Learn- ing a deep single image contrast enhancer from multi- exposure images,”IEEE TIP, vol. 27, no. 4, pp. 2049– 2062, 2018

    2018

  19. [27]

    The pascal visual object classes (voc) challenge,

    Mark Everingham, Luc Van Gool, Christopher KI Williams, John Winn, and Andrew Zisserman, “The pascal visual object classes (voc) challenge,”IJCV, vol. 88, pp. 303–338, 2010. CURVE: CLIP-UTILIZED REINFORCEMENT LEARNING FOR VISUAL IMAGE ENHANCEMENT VIA SIMPLE IMAGE PROCESSING Su...

    2010

  20. [28]

    B ´ezier-Curve Tone Adjustment (Sec

    SUPPLEMENTARY DETAILS 5.1. B ´ezier-Curve Tone Adjustment (Sec. 2.1) Tone adjustment maps the intensity of an input image to output image values, thereby changing the image contrast. Our method uses a cubic B´ezier curve to define this mapping function. A cubic B ´ezier curve ...

  21. [29]

    a good photo of cat

    EXTENDED QUALITA TIVE COMPARISONS 6.1. Multi-Exposure Image Dataset We show the comparison results on the SICE Part 2 dataset in Fig. 9 and Fig. 10. While our proposed method consistently maintains equivalent brightness levels across images with dif- ferent exposures, Zero-DCE...

Pith tools

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