Unpaired Learning for High Dynamic Range Image Tone Mapping

Vinker, Yael; Huberman-Spiegelglas, Inbar; Fattal, Raanan

doi:10.1109/iccv48922.2021.01439

Cited by 22 publications

(65 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…On LLE and EC tasks, LA-Net significantly outperforms the one-pathway model, which benefits from the high-frequency pathway with noise suppression and detail enhancement. However, The one-pathway model also achieves slightly better results with TM tasks and outperforms the recent method of Vinker et al [41]. This is reasonable considering that the TM task mainly focuses on dynamic range compression and the input HDR scenes contain fairly weak noises.…”

Section: Ablation Study and Parameter Analysissupporting

confidence: 59%

“…Table 4 lists the metrics obtained on the HDRPS dataset. Considering that the method of Vinker et al [41] outputs scaled images and image resizing affects the TMQI score [4], we also list the TMQI and BTMQI scores with the same resizing of the result images, denoted as LA-Net(resized), for a fair comparison. Note that difference exists between our reproduced scores and the ones in the original paper of Vinker et al, which could be caused by the different implementations of TMQI and BTMQI.…”

Section: Methodsmentioning

confidence: 99%

“…Inspired by the local adaptation mechanism of the biological visual system, some researchers have built models for TM based on the Retinex theory [27,28] or neural circuit in the retina [55]. Recent methods aimed to achieve TM with a deep generative adversarial network have also been reported [30,32,35,41].…”

Section: Related Workmentioning

confidence: 99%

“…There are multiple tasks in computer vision that are aimed at achieving light adaptation, such as low-light en- hancement [15,24,56,57], exposure correction [1,52,54], and high-dynamic-range (HDR) tone mapping [7,23,36,41]. Figure 1 shows three examples of these enhancement tasks.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Learning to Adapt to Light

Kaifu¹,

Cheng²,

Zhao³

et al. 2022

Preprint

View full text Add to dashboard Cite

Light adaptation or brightness correction is a key step in improving the contrast and visual appeal of an image. There are multiple light-related tasks (for example, low-light enhancement and exposure correction) and previous studies have mainly investigated these tasks individually. However, it is interesting to consider whether these light-related tasks can be executed by a unified model, especially considering that our visual system adapts to external light in such way. In this study, we propose a biologically inspired method to handle light-related image-enhancement tasks with a unified network (called LA-Net). First, a frequency-based decomposition module is designed to decouple the common and characteristic sub-problems of light-related tasks into two pathways. Then, a new module is built inspired by biological visual adaptation to achieve unified light adaptation in the low-frequency pathway. In addition, noise suppression or detail enhancement is achieved effectively in the high-frequency pathway regardless of the light levels. Extensive experiments on three tasks-low-light enhancement, exposure correction, and tone mapping-demonstrate that the proposed method almost obtains state-of-the-art performance compared with recent methods designed for these individual tasks.

show abstract

Section: Ablation Study and Parameter Analysissupporting

confidence: 59%

Section: Methodsmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Learning to Adapt to Light

Kaifu¹,

Cheng²,

Zhao³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Since local tone mapping requires fine-grained adjustment of each small patch in the scene, it is difficult to annotate it manually. Therefore, we use a pre-trained local tone mapping model in [37] to fulfill the task. Since the pre-trained tone mapping network was trained using daytime image, it is good for local adjustment, but fails to control the global brightness.…”

Section: Nr2r Datasetmentioning

confidence: 99%

Rendering Nighttime Image Via Cascaded Color and Brightness Compensation

Li¹,

Yi²,

Ma³

2022

Preprint

View full text Add to dashboard Cite

Image signal processing (ISP) is crucial for camera imaging, and neural networks (NN) solutions are extensively deployed for daytime scenes. The lack of sufficient nighttime image dataset and insights on nighttime illumination characteristics poses a great challenge for high-quality rendering using existing NN ISPs. To tackle it, we first built a high-resolution nighttime RAW-RGB (NR2R) dataset with white balance and tone mapping annotated by expert professionals. Meanwhile, to best capture the characteristics of nighttime illumination light sources, we develop the CBUnet, a two-stage NN ISP to cascade the compensation of color and brightness attributes. Experiments show that our method has better visual quality compared to traditional ISP pipeline, and is ranked at the second place in the NTIRE 2022 Night Photography Rendering Challenge [13] for two tracks by respective People's and Professional Photographer's choices. The code and relevant materials are avaiable on our website 1 .

show abstract

Towards Real-World HDRTV Reconstruction: A Data Synthesis-Based Approach

Cheng

Wang

et al. 2022

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Unpaired Learning for High Dynamic Range Image Tone Mapping

Cited by 22 publications

References 40 publications

Learning to Adapt to Light

Learning to Adapt to Light

Rendering Nighttime Image Via Cascaded Color and Brightness Compensation

Towards Real-World HDRTV Reconstruction: A Data Synthesis-Based Approach

Contact Info

Product

Resources

About