Stylization-Based Architecture for Fast Deep Exemplar Colorization

Xu, Zhongyou; Wang, Tingting; Fang, Faming; Yu, Shengkai; Zhang, Guixu

doi:10.1109/cvpr42600.2020.00938

Cited by 67 publications

(43 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Reference-based methods try to transfer the color statistics from the reference to the gray image using correspondences between the two based on low-level similarity measures [34], semantic features [6], or super-pixels [8,15]. Recent works [17,35,53] adopt deep neural network to improve the spatial correspondence and colorization results. Though these methods obtain remarkable results when suitable references are available, the procedure of finding references is time-consuming and challenging for automatic retrieval system [8].…”

Section: Related Workmentioning

confidence: 99%

“…However, the results are barely satisfactory and difficult to control (Detailed comparisons are provided in the supplementary material). For exemplar-based colorization methods [17,53], it is time-consuming and challenging to find large amounts of reference images with various styles. Different from previous methods, our method could achieve diverse and controllable colorization in a much easier and novel manner.…”

Section: Controllable Diverse Colorizationmentioning

confidence: 99%

“…Classic reference-based methods (e.g. [17,35,53]) require additional example color images as guidance. These methods attempt to match the relevant contents between exemplars and input gray images, and then transfer the color statistics from the reference to the gray one.…”

mentioning

confidence: 99%

See 2 more Smart Citations

Towards Vivid and Diverse Image Colorization with Generative Color Prior

Wang

Liu

et al. 2021

Preprint

View full text Add to dashboard Cite

Section: Related Workmentioning

confidence: 99%

Section: Controllable Diverse Colorizationmentioning

confidence: 99%

See 1 more Smart Citation

Towards Vivid and Diverse Image Colorization with Generative Color Prior

Wang

Liu

et al. 2021

Preprint

View full text Add to dashboard Cite

“…While the chrominance branch propagates color samples extracted from the reference to the entire image, the perceptual branch makes a prediction for areas left uncolored by the reference, purely based on dominant colors learned from the large-scale training set. This pattern was also used in the work of He et al [16], and Xu et al [49].…”

Section: Colorization Patterns and Learning Modelsmentioning

confidence: 99%

A Review and Analysis of The Existing Literature on Monochromatic Photography Colorization Using Deep Learning

Adascalitei

2021

Studia UBB Informatica

View full text Add to dashboard Cite

It is universally known that, through the process of colorization, one aims at converting a monochrome image into one of color, usually because it was taken by the limited technology of previous decades. Our work introduces the problem, summarizes the general deep learning solutions, and discusses the experimental results obtained from open-source repositories. Although the surveyed methods can be applied to other fields, solely the content of photography is being considered. Our contribution stands in the analysis of colorization in photography by examining used datasets and methodologies for evaluation, data processing activities, and the infrastructure demanded by these systems. We curated some of the most promising papers, published between 2016 and 2021, and centered our observations around software reliability, and key advancements in solutions employing Generative Adversarial Networks and Neural Networ

show abstract

“…He et al (Iizuka and Simo-Serra 2019) propose a similarity sub-net to compute the bidirectional similarity map between source and reference images. Considering the limitation of one-stage network, methods (Xu et al 2020;Zhang et al 2019a) design the coarse-to-fine net- However, the above methods are effortless to select incorrect referential information prone to produce visual artifacts, e.g., color shift and color patch. To alleviate the problems, we propose a novel PVCAttn module to more effectively aggregate information between source and reference images.…”

Section: Introductionmentioning

confidence: 99%

SCSNet: An Efficient Paradigm for Learning Simultaneously Image Colorization and Super-Resolution

Zhang¹,

Chen²,

Li³

et al. 2022

Preprint

View full text Add to dashboard Cite

In the practical application of restoring low-resolution grayscale images, we generally need to run three separate processes of image colorization, super-resolution, and dowssampling operation for the target device. However, this pipeline is redundant and inefficient for the independent processes, and some inner features could have been shared. Therefore, we present an efficient paradigm to perform Simultaneously Image Colorization and Super-resolution (SCS) and propose an end-to-end SCSNet to achieve this goal. The proposed method consists of two parts: colorization branch for learning color information that employs the proposed plug-and-play Pyramid Valve Cross Attention (PV-CAttn) module to aggregate feature maps between source and reference images; and super-resolution branch for integrating color and texture information to predict target images, which uses the designed Continuous Pixel Mapping (CPM) module to predict high-resolution images at continuous magnification. Furthermore, our SCSNet supports both automatic and referential modes that is more flexible for practical application. Abundant experiments demonstrate the superiority of our method for generating authentic images over state-of-theart methods, e.g., averagely decreasing FID by 1.8↓ and 5.1 ↓ compared with current best scores for automatic and referential modes, respectively, while owning fewer parameters (more than ×2↓) and faster running speed (more than ×3↑).

show abstract

Stylization-Based Architecture for Fast Deep Exemplar Colorization

Cited by 67 publications

References 16 publications

Towards Vivid and Diverse Image Colorization with Generative Color Prior

Towards Vivid and Diverse Image Colorization with Generative Color Prior

A Review and Analysis of The Existing Literature on Monochromatic Photography Colorization Using Deep Learning

SCSNet: An Efficient Paradigm for Learning Simultaneously Image Colorization and Super-Resolution

Contact Info

Product

Resources

About