Semantic Information Supplementary Pyramid Network for Dynamic Scene Deblurring
The algorithm in this paper is called semantic information supplementary pyramid network(SIS-net). We choose Generative Adversarial Network (GAN) as its fundamental model. SIS-net's generator imitates the feature pyramid network (FPN) structure to recycle features spanning across multiple recep...
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2020-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/9214402/ |
| _version_ | 1818876672067239936 |
|---|---|
| author | Yiming Liu Yifei Luo Wenzhuo Huang Ying Qiao Junhui Li Dahong Xu Duqiang Luo |
| author_facet | Yiming Liu Yifei Luo Wenzhuo Huang Ying Qiao Junhui Li Dahong Xu Duqiang Luo |
| author_sort | Yiming Liu |
| collection | DOAJ |
| description | The algorithm in this paper is called semantic information supplementary pyramid network(SIS-net). We choose Generative Adversarial Network (GAN) as its fundamental model. SIS-net's generator imitates the feature pyramid network (FPN) structure to recycle features spanning across multiple receptive scales to restore a sharp image. However, to solve the problem caused by the phenomenon of semantic dilution in the FPN network, we have innovatively designed a semantic information supplement (SIS) mechanism. SIS mechanism contains two essential components: semantic information storage box (info-box) and feature fusion expanding. In the process of feature fusion expanding, the semantic information features coming from the info-box is supplemented to make greater use of detailed clues. In addition, SIS-net uses the intermediate layer path to extract image features in a single time to obtain a multi-scale effect. The running speed of SIS-net has obvious advantages over other algorithms, and can basically complete real-time deblurring tasks. Extensive experiments show that our SIS-net achieves both qualitative and quantitative improvements against state-of-the-art methods. The code is available at https://github.com/yimingliu123/SIS-NET. |
| first_indexed | 2024-12-19T13:46:06Z |
| format | Article |
| id | doaj.art-1483137ede534fc498111efb24d12cfd |
| institution | Directory Open Access Journal |
| issn | 2169-3536 |
| language | English |
| last_indexed | 2024-12-19T13:46:06Z |
| publishDate | 2020-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Access |
| spelling | doaj.art-1483137ede534fc498111efb24d12cfd2022-12-21T20:18:51ZengIEEEIEEE Access2169-35362020-01-01818858718859910.1109/ACCESS.2020.30281579214402Semantic Information Supplementary Pyramid Network for Dynamic Scene DeblurringYiming Liu0https://orcid.org/0000-0002-0387-6329Yifei Luo1Wenzhuo Huang2https://orcid.org/0000-0002-7797-508XYing Qiao3https://orcid.org/0000-0002-5211-3109Junhui Li4https://orcid.org/0000-0003-3295-6050Dahong Xu5https://orcid.org/0000-0002-3265-8355Duqiang Luo6https://orcid.org/0000-0002-0885-3943Department of Bioinformatics, College of Life Science, Hebei University, Baoding, ChinaFaculty of Arts and Social Science, University of New South Wales, Sydney, NSW, AustraliaCollege of Information Science and Engineering, Hunan Normal University, Changsha, ChinaSchool of Arts and Science, Rutgers University, New Brunswick, NJ, USACollege of Information Science and Engineering, Hunan Normal University, Changsha, ChinaCollege of Information Science and Engineering, Hunan Normal University, Changsha, ChinaDepartment of Bioinformatics, College of Life Science, Hebei University, Baoding, ChinaThe algorithm in this paper is called semantic information supplementary pyramid network(SIS-net). We choose Generative Adversarial Network (GAN) as its fundamental model. SIS-net's generator imitates the feature pyramid network (FPN) structure to recycle features spanning across multiple receptive scales to restore a sharp image. However, to solve the problem caused by the phenomenon of semantic dilution in the FPN network, we have innovatively designed a semantic information supplement (SIS) mechanism. SIS mechanism contains two essential components: semantic information storage box (info-box) and feature fusion expanding. In the process of feature fusion expanding, the semantic information features coming from the info-box is supplemented to make greater use of detailed clues. In addition, SIS-net uses the intermediate layer path to extract image features in a single time to obtain a multi-scale effect. The running speed of SIS-net has obvious advantages over other algorithms, and can basically complete real-time deblurring tasks. Extensive experiments show that our SIS-net achieves both qualitative and quantitative improvements against state-of-the-art methods. The code is available at https://github.com/yimingliu123/SIS-NET.https://ieeexplore.ieee.org/document/9214402/Generative adversarial networkfeature pyramid networksemantic information |
| spellingShingle | Yiming Liu Yifei Luo Wenzhuo Huang Ying Qiao Junhui Li Dahong Xu Duqiang Luo Semantic Information Supplementary Pyramid Network for Dynamic Scene Deblurring IEEE Access Generative adversarial network feature pyramid network semantic information |
| title | Semantic Information Supplementary Pyramid Network for Dynamic Scene Deblurring |
| title_full | Semantic Information Supplementary Pyramid Network for Dynamic Scene Deblurring |
| title_fullStr | Semantic Information Supplementary Pyramid Network for Dynamic Scene Deblurring |
| title_full_unstemmed | Semantic Information Supplementary Pyramid Network for Dynamic Scene Deblurring |
| title_short | Semantic Information Supplementary Pyramid Network for Dynamic Scene Deblurring |
| title_sort | semantic information supplementary pyramid network for dynamic scene deblurring |
| topic | Generative adversarial network feature pyramid network semantic information |
| url | https://ieeexplore.ieee.org/document/9214402/ |
| work_keys_str_mv | AT yimingliu semanticinformationsupplementarypyramidnetworkfordynamicscenedeblurring AT yifeiluo semanticinformationsupplementarypyramidnetworkfordynamicscenedeblurring AT wenzhuohuang semanticinformationsupplementarypyramidnetworkfordynamicscenedeblurring AT yingqiao semanticinformationsupplementarypyramidnetworkfordynamicscenedeblurring AT junhuili semanticinformationsupplementarypyramidnetworkfordynamicscenedeblurring AT dahongxu semanticinformationsupplementarypyramidnetworkfordynamicscenedeblurring AT duqiangluo semanticinformationsupplementarypyramidnetworkfordynamicscenedeblurring |