Multi-Stage Progressive Image Restoration

Zamir, Syed Waqas; Arora, Aditya; Khan, Salman; Hayat, Munawar; Khan, Fahad Shahbaz; Yang, Ming–Hsuan; Shao, Ling

doi:10.1109/cvpr46437.2021.01458

Cited by 894 publications

(446 citation statements)

References 70 publications

Supporting

Mentioning

431

Contrasting

Order By: Relevance

“…9, only our method restores the contrast and image content approximate to the ground truth. In contrast, Input DCSFN [37]+ [32] DCSFN [37]+ [36] MPRNet [48]+ [32] MPRNet [48]+ [36] NLEDN [18]+ [32] NLEDN [18]+ [36] HR [21] Our GT Fig. 9.…”

Section: Results On Synthetic Rain-fog Datasetmentioning

confidence: 96%

“…Similar to [21], we combined several defogging (such as FFA [32] and DA [36]) and deraining (such as DCSFN [37], MPRNet [48] and NLEDN [18]) methods in different order to evaluate the derained performance on Rain-fog and HeavyRain [21] dateset. From defogging to deraining, we name as DFDR.…”

Section: Results On Synthetic Rain-fog Datasetmentioning

confidence: 99%

“…Fig. 11 shows the comparison results on several representative real-world scenarios with DerainNet [9], LPNet [9], MPRNet [48], SEMI [41], CycleDerain [42] and our method. We can see that, compared with our method, other methods can not remove the atmospheric veiling effect (as shown in the derained results in the third row of Fig.…”

Section: E Results On Real World Datasetmentioning

confidence: 99%

“…2) Comparison Methods and Evaluation Metrics: In this work, we qualitatively compare our proposed method with nine top-performing deraining methods. These baseline deraining methods are DCSFN [37], MPRNet [48], NLEDN [18], HR [21], DerainNet [9], SPANet [39], LPNet [11] SEMI [41] and CycleDerain [42]). Moreover, we also evaluate our method with some defogging methods, such as FFA [32], DA [36], Grid [28], DehazeNet [4], EPDN [33] and GCA [6].…”

Section: B Datasets Comparison Methods and Evaluation Metric 1) Datasetsmentioning

confidence: 99%

See 3 more Smart Citations

An Unsupervised Attentive-Adversarial Learning Framework for Single Image Deraining

Liu¹,

Jiang²,

Chen³

et al. 2022

Preprint

View full text Add to dashboard Cite

Single image deraining has been an important topic in low-level computer vision tasks. The atmospheric veiling effect (which is generated by rain accumulation, similar to fog) usually appears with the rain. Most deep learning-based single image deraining methods mainly focus on rain streak removal by disregarding this effect, which leads to low-quality deraining performance. In addition, these methods are trained only on synthetic data, hence they do not take into account real-world rainy images. To address the above issues, we propose a novel unsupervised attentive-adversarial learning framework (UALF) for single image deraining that trains on both synthetic and real rainy images while simultaneously capturing both rain streaks and rain accumulation features. UALF consists of a Rain-fog2Clean (R2C) transformation block and a Clean2Rainfog (C2R) transformation block. In R2C, to better characterize the rain-fog fusion feature and to achieve high-quality deraining performance, we employ an attention rain-fog feature extraction network (ARFE) to exploit the self-similarity of global and local rain-fog information by learning the spatial feature correlations. Moreover, to improve the transformation ability of C2R, we design a rain-fog feature decoupling and reorganization network (RFDR) by embedding a rainy image degradation model and a mixed discriminator to preserve richer texture details. Extensive experiments on benchmark rain-fog and rain datasets show that UALF outperforms state-of-the-art deraining methods. We also conduct defogging performance evaluation experiments to further demonstrate the effectiveness of UALF.

show abstract

Section: Results On Synthetic Rain-fog Datasetmentioning

confidence: 96%

Section: Results On Synthetic Rain-fog Datasetmentioning

confidence: 99%

Section: E Results On Real World Datasetmentioning

confidence: 99%

Section: B Datasets Comparison Methods and Evaluation Metric 1) Datasetsmentioning

confidence: 99%

See 2 more Smart Citations

An Unsupervised Attentive-Adversarial Learning Framework for Single Image Deraining

Liu¹,

Jiang²,

Chen³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Therefore, based on the traditional deblurring method, recovering a clear picture under the condition of a known blurred picture is a highly unsuitable problem. However, the blurring can be removed by applying different constraints to the blur kernel and the latent image, or using hand-made regularization items at the same time [9]. Unfortunately, these methods often lead to solving a non-convex optimization problem, so they cannot be well generalized to complex real examples [10].…”

Section: Introductionmentioning

confidence: 99%

A Survey on Image Deblurring

Li¹

2022

Preprint

View full text Add to dashboard Cite

With the improvement of social life quality and the real needs of daily work, images are more and more all around us. Image blurring due to camera shake, human movement, etc. has become the key to affecting image quality. How to remove image blur and restore clear image has gradually become an important research direction in the field of computer vision. After more than half a century of unremitting efforts, the majority of scientific and technological workers have made fruitful progress in image deblurring. This article reviews the work of image deblurring and specifically introduces more classic image deblurring methods, which is helpful to understand current research and look forward to future trends. This article reviews the traditional image deblurring methods and depth-represented image deblurring methods, and comprehensively classifies and introduces the corresponding technical methods. This review can provide some guidance for researchers in the field of image deblurring, and at the same time facilitate their subsequent study and research.Preprint. Under review.

show abstract

Single image deraining using multi‐stage and multi‐scale joint channel coordinate attention fusion network

Yang

Zhang

Cui

et al. 2022

Int J of Intelligent Sys

View full text Add to dashboard Cite

Rain streaks can seriously degrade the visual quality of an image and are detrimental to subsequent algorithms such as object detection and semantic segmentation. Therefore, removing rain streaks is a very important task. The deraining task has two main limitations: the first is to encode information about rain streaks in different densities and directions, the second is to keep the background details of the image while removing the rain streak. To address these limitations, we propose an effective algorithm, called multi-stage and multi-scale joint channel coordinate attention fusion network (MMAFN). We mainly propose a two-stage network structure, both of which use an encoderdecoder network to extract features. The first-stage network extracts coarse features and the second-stage network integrates the features of the former to further refine features. We design the joint channel coordinate attention block to encode features of rain streaks in different directions and densities. In addition, to better fuse features of different scales and enhance the generalization performance of the network, the inception attention branch block and the multi-level feature fusion block are designed. Extensive experiments substantiate the superiority of the proposed network

show abstract

Multi-Stage Progressive Image Restoration

Cited by 894 publications

References 70 publications

An Unsupervised Attentive-Adversarial Learning Framework for Single Image Deraining

An Unsupervised Attentive-Adversarial Learning Framework for Single Image Deraining

A Survey on Image Deblurring

Single image deraining using multi‐stage and multi‐scale joint channel coordinate attention fusion network

Contact Info

Product

Resources

About