A Multi-Strategy Framework for Coastal Waste Detection
In recent years, deep learning has been widely used in the field of coastal waste detection, with excellent results. However, there are difficulties in coastal waste detection such as, for example, detecting small objects and the low performance of the object detection model. To address these issues...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2022-09-01
|
| Series: | Journal of Marine Science and Engineering |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2077-1312/10/9/1330 |
| _version_ | 1827659973035819008 |
|---|---|
| author | Chengjuan Ren Sukhoon Lee Dae-Kyoo Kim Guangnan Zhang Dongwon Jeong |
| author_facet | Chengjuan Ren Sukhoon Lee Dae-Kyoo Kim Guangnan Zhang Dongwon Jeong |
| author_sort | Chengjuan Ren |
| collection | DOAJ |
| description | In recent years, deep learning has been widely used in the field of coastal waste detection, with excellent results. However, there are difficulties in coastal waste detection such as, for example, detecting small objects and the low performance of the object detection model. To address these issues, we propose the Multi-Strategy Deconvolution Single Shot Multibox Detector (MS-DSSD) based on DSSD. The method combines feature fusion, dense blocks, and focal loss into a state-of-the-art feed-forward network with an end-to-end training style. In the network, we employ feature fusion to import contextual information to boost the accuracy of small object detection. The dense blocks are constructed by a complex function of three concurrent operations, which can yield better feature descriptions. Then, focal loss is applied to address the class imbalance. Due to the lack of coastal waste datasets, data augmentation is designed to increase the amount of data, prevent overfitting of the model, and speed up convergence. Experimental results show that MS-DSSD513 obtains a higher mAP, of 82.2% and 84.1%, compared to the state-of-the-art object detection algorithms on PASCAL VOC2007 and our coastal waste dataset. The proposed new model is shown to be effective for small object detection and can facilitate the automatic detection of coastal waste management. |
| first_indexed | 2024-03-09T23:31:43Z |
| format | Article |
| id | doaj.art-f213b7f2897b45c78bdec0c0cc0b3cbe |
| institution | Directory Open Access Journal |
| issn | 2077-1312 |
| language | English |
| last_indexed | 2024-03-09T23:31:43Z |
| publishDate | 2022-09-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Journal of Marine Science and Engineering |
| spelling | doaj.art-f213b7f2897b45c78bdec0c0cc0b3cbe2023-11-23T17:08:37ZengMDPI AGJournal of Marine Science and Engineering2077-13122022-09-01109133010.3390/jmse10091330A Multi-Strategy Framework for Coastal Waste DetectionChengjuan Ren0Sukhoon Lee1Dae-Kyoo Kim2Guangnan Zhang3Dongwon Jeong4Guangdong Atv Academy for Performing Arts, Zhaoqing 526000, ChinaSoftware Convergence Engineering Department, Kunsan National University, Gunsan 54150, KoreaComputer Science and Engineering Department, Oakland University, Rochester, MI 48309, USADepartment of Computer Science, Baoji University of Arts and Science, Baoji 721000, ChinaSoftware Convergence Engineering Department, Kunsan National University, Gunsan 54150, KoreaIn recent years, deep learning has been widely used in the field of coastal waste detection, with excellent results. However, there are difficulties in coastal waste detection such as, for example, detecting small objects and the low performance of the object detection model. To address these issues, we propose the Multi-Strategy Deconvolution Single Shot Multibox Detector (MS-DSSD) based on DSSD. The method combines feature fusion, dense blocks, and focal loss into a state-of-the-art feed-forward network with an end-to-end training style. In the network, we employ feature fusion to import contextual information to boost the accuracy of small object detection. The dense blocks are constructed by a complex function of three concurrent operations, which can yield better feature descriptions. Then, focal loss is applied to address the class imbalance. Due to the lack of coastal waste datasets, data augmentation is designed to increase the amount of data, prevent overfitting of the model, and speed up convergence. Experimental results show that MS-DSSD513 obtains a higher mAP, of 82.2% and 84.1%, compared to the state-of-the-art object detection algorithms on PASCAL VOC2007 and our coastal waste dataset. The proposed new model is shown to be effective for small object detection and can facilitate the automatic detection of coastal waste management.https://www.mdpi.com/2077-1312/10/9/1330deep learningDeconvolution Single Shot Multibox Detectorwaste detection |
| spellingShingle | Chengjuan Ren Sukhoon Lee Dae-Kyoo Kim Guangnan Zhang Dongwon Jeong A Multi-Strategy Framework for Coastal Waste Detection Journal of Marine Science and Engineering deep learning Deconvolution Single Shot Multibox Detector waste detection |
| title | A Multi-Strategy Framework for Coastal Waste Detection |
| title_full | A Multi-Strategy Framework for Coastal Waste Detection |
| title_fullStr | A Multi-Strategy Framework for Coastal Waste Detection |
| title_full_unstemmed | A Multi-Strategy Framework for Coastal Waste Detection |
| title_short | A Multi-Strategy Framework for Coastal Waste Detection |
| title_sort | multi strategy framework for coastal waste detection |
| topic | deep learning Deconvolution Single Shot Multibox Detector waste detection |
| url | https://www.mdpi.com/2077-1312/10/9/1330 |
| work_keys_str_mv | AT chengjuanren amultistrategyframeworkforcoastalwastedetection AT sukhoonlee amultistrategyframeworkforcoastalwastedetection AT daekyookim amultistrategyframeworkforcoastalwastedetection AT guangnanzhang amultistrategyframeworkforcoastalwastedetection AT dongwonjeong amultistrategyframeworkforcoastalwastedetection AT chengjuanren multistrategyframeworkforcoastalwastedetection AT sukhoonlee multistrategyframeworkforcoastalwastedetection AT daekyookim multistrategyframeworkforcoastalwastedetection AT guangnanzhang multistrategyframeworkforcoastalwastedetection AT dongwonjeong multistrategyframeworkforcoastalwastedetection |