Dynamic Scene Deblurring With Parameter Selective Sharing and Nested Skip Connections

By Hongyun Gao, Xin Tao, Xiaoyong Shen, Jiaya Jia. Computer Vision and Pattern Recognition, 2019

414 citations recorded by the source corpus.

Dynamic Scene deblurring is a challenging low-level vision task where spatially variant blur is caused by many factors, e.g., camera shake and object motion. Recent study has made significant progress. Compared with the parameter independence scheme [19] and parameter sharing scheme [33], we develop the general principle for constraining the deblurring network structure by proposing the generic and effective selective sharing scheme. Inside the subnetwork of each scale, we propose a nested skip connection structure for the nonlinear transformation modules to replace stacked convolution layers or residual blocks. Besides, we build a new large dataset of blurred/sharp image pairs towards better restoration quality. Comprehensive experimental results show that our parameter selective sharing scheme, nested skip connection structure, and the new dataset are all significant to set a new state-of-the-art in dynamic scene deblurring.

Subfields: Computer Vision and Pattern Recognition, Media Technology

View Paper: https://www.semanticscholar.org/paper/a58468dace1b8654f4ab5a92411ad3c76998c280

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Media Technology

Citations in the corpus

Neural Blind Deconvolution Using Deep Priors (2019, methodology): “Given the training set, deep CNNs can either be used to extract features to facilitate blur kernel estimation [1, 38], or be deployed to learn the direct mapping to clean image for motion deblurring [7,25,43,51].”
Spatially-Attentive Patch-Hierarchical Network for Adaptive Motion Deblurring (2020, neutral): “Method [24] [23] [5] [3] [13] [8] [20] [27] [2] [26] [9] Ours(a) Ours(b)”
Stripformer: Strip Transformer for Fast Image Deblurring (2022, methodology): “The deblurred results from left to right are produced by DSD [6], DMPHN [9], MTRNN [10], MIMO [42], MPRNet [33] and our method, respectively. fine-grained information like facial expressions on the HIDE dataset.”
Deep Image Deblurring: A Survey (2022, methodology): “…multi-scale architecture (Gao et al. 2019; Purohit et al. 2019; Tao et al. 2018), improving deblurring at different scales, e . g ., using nested skip connections in Gao et al. (2019), or increasing the connection of networks across different scales, e . g ., recurrent layers in Tao et al. (2018).”
Artificial intelligence in the creative industries: a review (2020, neutral): “ture. The high-level features of each iteration are linked in a recurrent manner, leading to a recursive process of learning sharp images from blurred ones. Nested skip connections were introduced in [111], where feature maps from multiple convolution layers are merged before applying them to the ne”
Pre-Trained Image Processing Transformer (2020, methodology): “Method MSCNN [54] SRN [67] DSD [30] DeblurGANv2 [44] DMPHN [84] LEBMD [40] EDSD [81]”
Deblurring via Stochastic Refinement (2021, methodology): “Alternatively, many recent methods adopt an end-to-end approach where a deep neural network is trained to directly produce a point estimate [8, 11, 14, 21, 35, 36, 50, 54, 55, 64, 66, 67, 73].”