Underwater Target Detection Lightweight Algorithm Based on Multi-Scale Feature Fusion

被引:19
作者
Chen, Liang [1 ]
Yang, Yuyi [1 ]
Wang, Zhenheng [1 ]
Zhang, Jian [1 ]
Zhou, Shaowu [1 ]
Wu, Lianghong [1 ]
机构
[1] Hunan Univ Sci & Technol, Sch Informat & Elect Engn, Xiangtan 411201, Peoples R China
基金
中国国家自然科学基金;
关键词
underwater target detection; multi-scale fusion; transformer; YOLOv5; lightweight;
D O I
10.3390/jmse11020320
中图分类号
U6 [水路运输]; P75 [海洋工程];
学科分类号
0814 ; 081505 ; 0824 ; 082401 ;
摘要
The performance of underwater target detection algorithms is affected by poor imaging quality in underwater environments. Due to the arithmetic power limitation of underwater devices, existing deep learning networks are unable to provide efficient detection processes with high detection accuracy. Lightweight CNN models have been actively applied for underwater environment detection, yet their lite feature fusion networks cannot provide effective fusion effects and reduce the detection accuracy. In this paper, a lightweight algorithm based on multi-scale feature fusion was proposed, with the model parameters greatly reduced, improving the target detection accuracy. The forward propagation memory overhead is reduced by using multi-scale shared convolutional kernels and pooling operations to co-construct the query matrix in the Tansformer encoding stage. Then, the feature fusion path is optimized in order to enhance the connection of multi-scale features. A multiscale feature adaptive fusion strategy is used to enhance the detection performance and reduce the dependence on the complex feature extraction network. The feature extraction network is also reparameterized to simplify the operation. Using the UPRC offshore dataset for validation, the study results have demonstrated that the statistical mAP metrics validate the detection accuracy. Compared with SSD, RetinaNet and YOLOv5-s improved by 13%, 8.6%, and 0.8%, while the number of parameters decreased by 76.09%, 89.74%, and 87.67%. In addition, compared with the YOLOv5-lite model algorithm with the same parameter volume, the mAP is improved by 3.8%, which verifies the accuracy and efficiency of the algorithm in this paper.
引用
收藏
页数:17
相关论文
共 30 条
[1]   YOLO-Fish: A robust fish detection model to detect fish in realistic underwater environment [J].
Al Muksit, Abdullah ;
Hasan, Fakhrul ;
Emon, Md. Fahad Hasan Bhuiyan ;
Haque, Md Rakibul ;
Anwary, Arif Reza ;
Shatabda, Swakkhar .
ECOLOGICAL INFORMATICS, 2022, 72
[2]   SWIPENET: Object detection in noisy underwater scenes [J].
Chen, Long ;
Zhou, Feixiang ;
Wang, Shengke ;
Dong, Junyu ;
Li, Ning ;
Ma, Haiping ;
Wang, Xin ;
Zhou, Huiyu .
PATTERN RECOGNITION, 2022, 132
[3]  
Chen Y., 2022, P IEEECVF C COMPUTER, V2, P6
[4]   RepVGG: Making VGG-style ConvNets Great Again [J].
Ding, Xiaohan ;
Zhang, Xiangyu ;
Ma, Ningning ;
Han, Jungong ;
Ding, Guiguang ;
Sun, Jian .
2021 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, CVPR 2021, 2021, :13728-13737
[5]  
Dosovitskiy A., 2021, arXiv
[6]   A novel sonar target detection and classification algorithm [J].
Fan, Xinnan ;
Lu, Liang ;
Shi, Pengfei ;
Zhang, Xuewu .
MULTIMEDIA TOOLS AND APPLICATIONS, 2022, 81 (07) :10091-10106
[7]   Delving into the Estimation Shift of Batch Normalization in a Network [J].
Huang, Lei ;
Zhou, Yi ;
Wang, Tian ;
Luo, Jic ;
Liu, Xianglong .
2022 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2022), 2022, :753-762
[8]  
Jiang Y., 2021, P INT C LEARNING REP
[9]  
Liu S., 2019, arXiv
[10]   Path Aggregation Network for Instance Segmentation [J].
Liu, Shu ;
Qi, Lu ;
Qin, Haifang ;
Shi, Jianping ;
Jia, Jiaya .
2018 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2018, :8759-8768