[1]王德兴,何勇,袁红春.基于YOLOv8-BAN模型的水下生物目标检测方法[J].江苏农业学报,2025,(01):101-111.[doi:doi:10.3969/j.issn.1000-4440.2025.01.012]
　WANG Dexing,HE Yong,YUAN Hongchun.Underwater biological target detection method based on YOLOv8-BAN model[J].,2025,(01):101-111.[doi:doi:10.3969/j.issn.1000-4440.2025.01.012]
点击复制

基于YOLOv8-BAN模型的水下生物目标检测方法()
分享到：

江苏农业学报[ISSN:1006-6977/CN:61-1281/TN]

卷:
期数:
2025年01期
页码:
101-111
栏目:
农业信息工程
出版日期:
2025-01-31

文章信息/Info

Title:
Underwater biological target detection method based on YOLOv8-BAN model
作者:
王德兴; 何勇; 袁红春
(上海海洋大学信息学院,上海201306)
Author(s):
WANG Dexing; HE Yong; YUAN Hongchun
(College of Information Technology, Shanghai Ocean University, Shanghai 201306, China)
关键词:
水下生物; YOLOv8m; 深度学习; 小目标检测
Keywords:
underwater organisms; YOLOv8m; deep learning; small target detection
分类号:
TP391
DOI:
doi:10.3969/j.issn.1000-4440.2025.01.012
文献标志码:
A
摘要:
水下目标检测技术对于自动化水下捕捞至关重要,可有效推动渔业的智能化发展。针对水下图像质量较差和小目标水下生物聚集导致漏检、误检等问题,本研究提出了一种基于改进YOLOv8m模型的水下生物目标检测模型——YOLOv8-BAN。该模型首先在骨干网络中嵌入双向路由自注意力机制,以增强网络的特征提取能力；其次在颈部结合自适应特征融合模块,优化特征融合效果,增强了模型对多尺度目标的检测能力；最后设计了一种小目标损失函数,通过精确标签分配进一步提升了水下生物小目标的检测精度。在URPC2018和Brackish数据集上的测试结果显示,YOLOv8-BAN模型的平均检测精度分别达到86.9%和98.6%,较YOLOv8m分别提高了3.5个百分点和3.3个百分点；与其他6种模型相比,YOLOv8-BAN模型具有更高的检测精度和较快的检测速度。本研究结果可为水下机器人进行水产捕捞作业提供了技术支持。
Abstract:
Underwater target detection technology is crucial for the automation of underwater fishing, which can effectively promote the intelligent development of the fishing industry. Aiming at the problems of poor underwater image quality and missed and false detections caused by the aggregation of small target underwater organisms, this study proposed an underwater biological target detection method based on improved YOLOv8m model, namely YOLOv8-BAN. The model first embedded a bidirectional routing self-attention mechanism in the backbone network to enhance the network’s feature extraction capability. Secondly, the adaptive feature fusion module was combined in the neck to optimize feature fusion effects, enhancing the model’s detection capability for multi-scale targets. Finally, a small target loss function was designed to further improve the detection accuracy of small targets through precise label assignment. Experimental results on the URPC2018 and Brackish datasets showed that the average detection accuracy of YOLOv8-BAN model reached 86.9% and 98.6% respectively, which was 3.5 percentage points and 3.3 percentage points higher than that of YOLOv8m model. Compared with the other six models, the YOLOv8-BAN model had higher detection accuracy and faster detection speed. The results of this study can provide technical support for underwater robots to carry out aquaculture fishing operations.

参考文献/References:

[1]FAYAZ S, PARAH S A, QURESHI G J, et al. Underwater object detection:architectures and algorithms-a comprehensive review[J]. Multimedia Tools and Applications,2022,81(1):20871-20916.
[2]许裕良,杜江辉,雷泽宇,等. 水下机器人在渔业中的应用现状与关键技术综述[J]. 机器人,2023,45(1):110-128.
[3]XU S B, ZHANG M H, SONG W, et al. A systematic review and analysis of deep learning-based underwater object detection[J]. Neurocomputing,2023,527:204-232.
[4]REN S Q, HE K M, GIRSHICK R, et al. Faster R-CNN: towards real-time object detection with region proposal networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2017,39(6):1137-1149.
[5]袁红春,张硕. 基于Faster R-CNN和图像增强的水下鱼类目标检测方法[J]. 大连海洋大学学报,2020,35(4):612-619.
[6]LIU J, LIU S, XU S J, et al. Two-stage underwater object detection network using swin transformer[J]. IEEE Access,2022,10:117235-117247.
[7]LIN W H, ZHONG J X, LIU S, et al. Roimix: proposal-fusion among multiple images for underwater object detection[C]. Barcelona:ICASSP,2020.
[8]SHI P, XU X, NI J, et al. Underwater biological detection algorithm based on improved faster-RCNN[J]. Water,2021,13(17):2420.
[9]REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once:unified,real-time object detection[C]. Las Vegas:IEEE,2016.
[10]REDMON J, FARHADI A. YOLO9000:better,faster,stronger[C]. Honolulu:IEEE,2017.
[11]REDMON J, FARHADI A. Yolov3:an incremental improvement[C]. Salt Lake City:CVPR,2018.
[12]BOCHKOVSKIY A, WANG C Y, LIAO H Y M, et al. Yolov4:optimal speed and accuracy of object detection[C]. Seattle:CVPR,2020.
[13]GUO T, WEI Y, SHAO H, et al. Research on underwater target detection method based on improved MSRCP and YOLOv3[C]. Nashville:IEEE,2021.
[14]CHEN L Y, ZHENG M C, DUAN S Q, et al. Underwater target recognition based on improved YOLOv4 neural network[J]. Electronics,2021,10(14):1634.
[15]LEI F, TANG F, LI S. Underwater target detection algorithm based on improved YOLOv5[J]. Journal of Marine Science and Engineering,2022,10(3):310.
[16]翟先一,魏鸿磊,韩美奇,等. 基于改进YOLO卷积神经网络的水下海参检测[J]. 江苏农业学报,2023,39(7):1543-1553.
[17]SUN Y, ZHENG W X, DU X, et al. Underwater small target detection based on YOLOX combined with mobileViT and double coordinate attention[J]. Journal of Marine Science and Engineering,2023,11(6):1178.
[18]YI W G, WANG B. Research on underwater small target detection algorithm based on improved YOLOv7[J]. IEEE Access,2023,11:66818-66827.
[19]ZHU L, WANG X, KE Z, et al. BiFormer:vision transformer with Bi-level routing attention[C]. Vancouver:IEEE,2023.
[20]REN S, ZHOU D, HE S, et al. Shunted self-attention via multi-scale token aggregation[C]. New Orleans:IEEE,2022.
[21]ZHENG Z, WANG P, LIU W, et al. Distance-IoU loss:faster and better learning for bounding box regression[C]. New York:AAAI,2020.
[22]XU C, WANG J W, YANG W, et al. Detecting tiny objects in aerial images:a normalized Wasserstein distance and a new benchmark[J]. ISPRS Journal of Photogrammetry and Remote Sensing,2022,190:79-93.
[23]ARJOVSKY M, CHINTALA S, BOTTOU L. Wasserstein generative adversarial networks[C]. Sydney:ICML,2017.
[24]WANG D D, HE D J. Channel pruned YOLO V5s-based deep learning approach for rapid and accurate apple fruitlet detection before fruit thinning[J]. Biosystems Engineering,2021,210:271-281.
[25]HE Q, XU A, YE Z, et al. Object detection based on lightweight YOLOX for autonomous driving[J]. Sensors,2023,23(17):7596.
[26]LI Y H, MAO H Z, GIRSHICK R, et al. Exploring plain vision transformer backbones for object detection[C]. Tel Aviv:ECCV,2022.
[27]WANG C Y, BOCHKOVSKIY A, LIAO H Y M. YOLOv7:trainable bag-of-freebies sets new state-of-the-art for real-time object detectors[C]. Vancouver:IEEE,2023.

备注/Memo

备注/Memo:
收稿日期：2024-04-03基金项目：国家自然科学基金项目(41776142)作者简介：王德兴(1971-),男,河北保定人,博士,副教授,研究方向为人工智能、模式识别和数据挖掘等。(E-mail)dxwang@shou.edu.cn通讯作者：何勇,(E-mail)2850035542@qq.com
更新日期/Last Update: 2025-02-28