2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) 2020
DOI: 10.1109/cvprw50498.2020.00237
|View full text |Cite
|
Sign up to set email alerts
|

High-Resolution Dual-Stage Multi-Level Feature Aggregation for Single Image and Video Deblurring

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
2

Citation Types

0
12
0

Year Published

2020
2020
2024
2024

Publication Types

Select...
3
3
2

Relationship

0
8

Authors

Journals

citations
Cited by 28 publications
(12 citation statements)
references
References 24 publications
0
12
0
Order By: Relevance
“…Therefore, they selectively chose the share module at different scales. Some works divide the deblurring process into two steps [19]- [22]. Chen [19] removed the blur effect in the Fourier domain and then designed a network for denoising.…”
Section: B Learning-based Methodsmentioning
confidence: 99%
See 2 more Smart Citations
“…Therefore, they selectively chose the share module at different scales. Some works divide the deblurring process into two steps [19]- [22]. Chen [19] removed the blur effect in the Fourier domain and then designed a network for denoising.…”
Section: B Learning-based Methodsmentioning
confidence: 99%
“…Brehm [21] used a two-step strategy for video deblurring; single image deblurring is performed first, and then a temporal fusion is applied for better performance. Gai [22] fused multiple blurry frames in two steps; a fused image is created by the first network before it is sent to the second network for deblurring. Although those methods divide the deblurring process into two steps for various purposes, none of them solve the inconsistency of the distribution between the restored and latent images.…”
Section: B Learning-based Methodsmentioning
confidence: 99%
See 1 more Smart Citation
“…Learning based methods, accentuated by deep learning, have achieved stateof-the-art performance on a variety of image restoration tasks including deblurring [5,41,47], dehazing [3,6,44], denoising [1,66,68], deraining [6,30] and image enhancement [8,13]. However, deep learning techniques face two main drawbacks with regard to UDC imaging systems.…”
Section: Introductionmentioning
confidence: 99%
“…Hierarchical dense residual learning is proposed via multi-level dense connections and multi-level residual connections.To implement multilevel dense connection, 1×1 convolution layers are inserted as the first and the last layers of MDCG and MDCB modules[41], reducing the number of feature maps. Inspired by[7], multi-scale feature extraction modules are used without reducing spatial resolution. To achieve the implementation principle, MDCB[41] and Laplacian attention[4] modules are used with modifications.…”
mentioning
confidence: 99%