Joint channel-spatial attention network for super-resolution image quality assessment

Zhang, Tingyue; Zhang, Kaibing; Xiao, Chuan; Xiong, Zenggang; Lü, Jian

doi:10.1007/s10489-022-03338-1

Cited by 12 publications

(7 citation statements)

References 41 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Ref. [23] proposed a method to combine channel attention with spatial attention; Ref. [20]exploited spatial attention, Ref.…”

Section: B Image Quality Assessment Based On Attention Mechanismmentioning

confidence: 99%

“…Since different attention mechanisms can pay attention to different regions of an image, it is inaccurate to rely on only one attention mechanism to extract features. Some researchers fused channel attention, spatial attention, and kernel attention to solve image processing problems [23], [24], but most of them used the form of simple cascade of attention modules without paying attention to the effective fusion of three attention mechanisms.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A Novel Multi-attention Fusion Convolution Neural Network for Blind Image Quality Assessment

Wang

Zhang

et al. 2022

Preprint

View full text Add to dashboard Cite

As an underlying computer vision task, blind image quality assessment plays an important role in image evaluation. According to the Human Vision System (HVS), when observing an image, people's visual attention will be more sensitive to the region of their concern, which shows that attention mechanism is very helpful for image quality assessment. Therefore, in blind image quality assessment, many attention mechanisms were used to extract the features of concerned regions. However, most of the existing methods only cascaded multiple attention mechanisms to extract features, which greatly weakened the effect of the fusion of multiple attention mechanisms. Based on this, we propose a multi-attention fusion convolution neural network (MAFCN), which can extract the features of interest region from different aspects. In this network, a multi-attention fusion unit (MAFU) is proposed which can adaptively fuse three types of attention mechanisms and give full play to the ability of each attention mechanism to extract features. In addition, skip connections are added to the network to supplement local information and missing details. Experimental results show that MAFCN has better performance and strong generalization ability without wasting resources.

show abstract

“…Ref. [23] proposed a method to combine channel attention with spatial attention; Ref. [20]exploited spatial attention, Ref.…”

Section: B Image Quality Assessment Based On Attention Mechanismmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

A Novel Multi-attention Fusion Convolution Neural Network for Blind Image Quality Assessment

Wang

Zhang

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Lu et al [13] deepen the depth of IQA models, while Zhao et al [14] and Liu et al [15] increase the width of IQA models, i.e., adopting dual-branch architectures. Besides, Zhang et al [16] introduce channel-spatial attention mechanisms into IQA models. However, these methods are still performance limited since they only consider the image features while ignoring the available quality-aware priori, i.e., the scale information.…”

Section: A Super-resolution Image Quality Assessmentmentioning

confidence: 99%

“…To verify the effectiveness of the proposed framework, we apply it to five representative blind IQA methods: CN-NIQA [24], ResNet50 [25] based IQA metric, HyperIQA [21], JCSAN [16], and DeepSRQ [10]. The single-dataset and crossdataset comparisons between the SGH-enhanced IQA metrics with the original ones are presented in Table III and Table IV, respectively.…”

Section: Effectiveness Of Scale Guided Hypernetwork Frameworkmentioning

confidence: 99%

“…With the success of the first deep learning-based blind SR IQA measure, researchers shift their focus from feature engineering to neural architecture engineering. In the past few years, various neural architecture strategies including deeper network architecture [13], dual-branch framework [14], [15], and visual attention mechanisms [16], have been proposed to further improve the performance.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation