Residual Attention Network for Image Classification

Wang, Fei; Jiang, Mengqing; Qian, Chen; Yang, Shuo; Cheng, Li; Zhang, Honggang; Wang, Xiaogang; Tang, Xiaoou

doi:10.1109/cvpr.2017.683

Cited by 3,209 publications

(1,888 citation statements)

References 54 publications

Supporting

Mentioning

1,761

Contrasting

Unclassified

Order By: Relevance

“…Second, by improving the network design and loss function to prevent the removal of subtle true‐positive signals. The loss function might be further constrained to prevent removal of the true high‐intensity signals and the network design might be further advanced to learn more complex spatial patterns such as using more residual connections or different modules such as the attention mechanism …”

Section: Discussionmentioning

confidence: 99%

Data‐driven synthetic MRI FLAIR artifact correction via deep neural network

Ryu

Nam

Gho³

et al. 2019

Magnetic Resonance Imaging

View full text Add to dashboard Cite

Background FLAIR (fluid attenuated inversion recovery) imaging via synthetic MRI methods leads to artifacts in the brain, which can cause diagnostic limitations. The main sources of the artifacts are attributed to the partial volume effect and flow, which are difficult to correct by analytical modeling. In this study, a deep learning (DL)‐based synthetic FLAIR method was developed, which does not require analytical modeling of the signal. Purpose To correct artifacts in synthetic FLAIR using a DL method. Study Type Retrospective. Subjects A total of 80 subjects with clinical indications (60.6 ± 16.7 years, 38 males, 42 females) were divided into three groups: a training set (56 subjects, 62.1 ± 14.8 years, 25 males, 31 females), a validation set (1 subject, 62 years, male), and the testing set (23 subjects, 57.3 ± 20.4 years, 13 males, 10 females). Field Strength/Sequence 3 T MRI using a multiple‐dynamic multiple‐echo acquisition (MDME) sequence for synthetic MRI and a conventional FLAIR sequence. Assessment Normalized root mean square (NRMSE) and structural similarity (SSIM) were computed for uncorrected synthetic FLAIR and DL‐corrected FLAIR. In addition, three neuroradiologists scored the three FLAIR datasets blindly, evaluating image quality and artifacts for sulci/periventricular and intraventricular/cistern space regions. Statistical Tests Pairwise Student's t‐tests and a Wilcoxon test were performed. Results For quantitative assessment, NRMSE improved from 4.2% to 2.9% (P < 0.0001) and SSIM improved from 0.85 to 0.93 (P < 0.0001). Additionally, NRMSE values significantly improved from 1.58% to 1.26% (P < 0.001), 3.1% to 1.5% (P < 0.0001), and 2.7% to 1.4% (P < 0.0001) in white matter, gray matter, and cerebral spinal fluid (CSF) regions, respectively, when using DL‐corrected FLAIR. For qualitative assessment, DL correction achieved improved overall quality, fewer artifacts in sulci and periventricular regions, and in intraventricular and cistern space regions. Data Conclusion The DL approach provides a promising method to correct artifacts in synthetic FLAIR. Level of Evidence: 4 Technical Efficacy: Stage 1 J. Magn. Reson. Imaging 2019;50:1413–1423.

show abstract

Section: Discussionmentioning

confidence: 99%

Data‐driven synthetic MRI FLAIR artifact correction via deep neural network

Ryu

Nam

Gho³

et al. 2019

Magnetic Resonance Imaging

View full text Add to dashboard Cite

show abstract

“…This function helps to filter out unimportant information and improve the efficiency of information processing. In deep learning, this idea has been explicitly simulated as “Attention.” Attention mechanisms are widely used in natural language processing, computer vision, and other fields.…”

Section: Related Workmentioning

confidence: 99%

A segmentation network with multiattention and its application to SAR image analysis

Long

Mabu

2020

IEEJ Transactions Elec Engng

View full text Add to dashboard Cite

Image segmentation plays an important role in image understanding and region‐based applications. Many image segmentation algorithms have been proposed, but in this paper, we enhance the segmentation performance of deep learning using attention models that extract important features from the target images. The structure of the segmentation network is an encoder–decoder model that can combine position features and channel features, where the attention mechanisms refine both position and channel features. In the experiments, the proposed method is applied to satellite image analysis, where synthetic aperture radar images are analyzed to detect landslide areas after heavy rain occurred. The experimental results show that the proposed method obtains higher segmentation accuracy comparing with some conventional methods. © 2019 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.

show abstract

“…Attention, or so called saliency detection, has been shown to play an important role in a wide variety of computer vision and robotics tasks [4,6,[23][24][25]. Despite their different application scenarios, such works utilize a neural network to learn to automatically locate task-relevant regions.…”

Section: B Attention Model For Place Recognitionmentioning

confidence: 99%

“…There have been attempts in utilizing attention in many tasks, such as image classification [4], image retrieval [5] or segmentation [6], while attention can also play an important role in visual place recognition [7]. Visual cues that are relevant to place recognition are generally not uniformly distributed across an image, therefore focusing on important regions, as opposed to irrelevant or confusion areas, is key to improve the place recognition performance.…”

mentioning

confidence: 99%