Multibranch Attention Mechanism Based on Channel and Spatial Attention Fusion
Recently, it has been demonstrated that the performance of an object detection network can be improved by embedding an attention module into it. In this work, we propose a lightweight and effective attention mechanism named multibranch attention (M3Att). For the input feature map, our M3Att first us...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-11-01
|
| Series: | Mathematics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2227-7390/10/21/4150 |
| _version_ | 1797467334128959488 |
|---|---|
| author | Guojun Mao Guanyi Liao Hengliang Zhu Bo Sun |
| author_facet | Guojun Mao Guanyi Liao Hengliang Zhu Bo Sun |
| author_sort | Guojun Mao |
| collection | DOAJ |
| description | Recently, it has been demonstrated that the performance of an object detection network can be improved by embedding an attention module into it. In this work, we propose a lightweight and effective attention mechanism named multibranch attention (M3Att). For the input feature map, our M3Att first uses the grouped convolutional layer with a pyramid structure for feature extraction, and then calculates channel attention and spatial attention simultaneously and fuses them to obtain more complementary features. It is a “plug and play” module that can be easily added to the object detection network and significantly improves the performance of the object detection network with a small increase in parameters. We demonstrate the effectiveness of M3Att on various challenging object detection tasks, including PASCAL VOC2007, PASCAL VOC2012, KITTI, and Zhanjiang Underwater Robot Competition. The experimental results show that this method dramatically improves the object detection effect, especially for the PASCAL VOC2007, and the mapping index of the original network increased by 4.93% when embedded in the YOLOV4 (You Only Look Once v4) network. |
| first_indexed | 2024-03-09T18:52:08Z |
| format | Article |
| id | doaj.art-aaecb70bc73545238926822fd1274ec9 |
| institution | Directory Open Access Journal |
| issn | 2227-7390 |
| language | English |
| last_indexed | 2024-03-09T18:52:08Z |
| publishDate | 2022-11-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Mathematics |
| spelling | doaj.art-aaecb70bc73545238926822fd1274ec92023-11-24T05:45:39ZengMDPI AGMathematics2227-73902022-11-011021415010.3390/math10214150Multibranch Attention Mechanism Based on Channel and Spatial Attention FusionGuojun Mao0Guanyi Liao1Hengliang Zhu2Bo Sun3Fujian Provincial Key Laboratory of Big Data Mining and Applications, Fujian University of Technology, Fuzhou 350118, ChinaSchool of Computer and Mathematics, Fujian University of Technology, Fuzhou 350118, ChinaSchool of Computer and Mathematics, Fujian University of Technology, Fuzhou 350118, ChinaSchool of Computer and Mathematics, Fujian University of Technology, Fuzhou 350118, ChinaRecently, it has been demonstrated that the performance of an object detection network can be improved by embedding an attention module into it. In this work, we propose a lightweight and effective attention mechanism named multibranch attention (M3Att). For the input feature map, our M3Att first uses the grouped convolutional layer with a pyramid structure for feature extraction, and then calculates channel attention and spatial attention simultaneously and fuses them to obtain more complementary features. It is a “plug and play” module that can be easily added to the object detection network and significantly improves the performance of the object detection network with a small increase in parameters. We demonstrate the effectiveness of M3Att on various challenging object detection tasks, including PASCAL VOC2007, PASCAL VOC2012, KITTI, and Zhanjiang Underwater Robot Competition. The experimental results show that this method dramatically improves the object detection effect, especially for the PASCAL VOC2007, and the mapping index of the original network increased by 4.93% when embedded in the YOLOV4 (You Only Look Once v4) network.https://www.mdpi.com/2227-7390/10/21/4150object detectionmultiscale modulespatial attentionchannel attentionspatial attentionmultibranch structure |
| spellingShingle | Guojun Mao Guanyi Liao Hengliang Zhu Bo Sun Multibranch Attention Mechanism Based on Channel and Spatial Attention Fusion Mathematics object detection multiscale module spatial attention channel attention spatial attention multibranch structure |
| title | Multibranch Attention Mechanism Based on Channel and Spatial Attention Fusion |
| title_full | Multibranch Attention Mechanism Based on Channel and Spatial Attention Fusion |
| title_fullStr | Multibranch Attention Mechanism Based on Channel and Spatial Attention Fusion |
| title_full_unstemmed | Multibranch Attention Mechanism Based on Channel and Spatial Attention Fusion |
| title_short | Multibranch Attention Mechanism Based on Channel and Spatial Attention Fusion |
| title_sort | multibranch attention mechanism based on channel and spatial attention fusion |
| topic | object detection multiscale module spatial attention channel attention spatial attention multibranch structure |
| url | https://www.mdpi.com/2227-7390/10/21/4150 |
| work_keys_str_mv | AT guojunmao multibranchattentionmechanismbasedonchannelandspatialattentionfusion AT guanyiliao multibranchattentionmechanismbasedonchannelandspatialattentionfusion AT hengliangzhu multibranchattentionmechanismbasedonchannelandspatialattentionfusion AT bosun multibranchattentionmechanismbasedonchannelandspatialattentionfusion |