CAGE-YOLO： 基于遥感影像的水产养殖网箱高密度小目标检测模型

doi:10.12133/j.smartag.SA202508023

摘要/Abstract

摘要：

【目的/意义】 针对复杂背景下密集且小尺度海水网箱目标检测困难的问题，本研究构建专用数据集并设计具有针对性的检测模型，以提升在实际养殖管理中的识别精度与鲁棒性。 【方法】 基于来自澳大利亚、加拿大、智利、克罗地亚、希腊、中国及法罗群岛七个具有代表性的海水养殖区域的高分辨率遥感影像，构建了海水网箱数据集；提出了基于 YOLOv5的Cage-YOLO 深度学习检测模型，用于密集小目标网箱的自动识别。首先，引入自适应密集感知算法，自动选择并生成反映小型网箱密集分布特征的特征图；其次，集成改进的快速空间金字塔池化模块，有效减少背景噪声干扰并增强全局特征提取能力；最后，加入混合注意力模块，进一步提升模型对密集小目标的感知能力。 【结果和讨论】 实验结果表明，Cage-YOLO在精度、召回率及平均精度方面分别较原始YOLOv5提升了5.6、21.8和17.4个百分点。模型体积保持为16.9 MB，兼具优越性能与良好的部署优势。 【结论】 本研究为密集小目标检测提供了一种有效的新方法，并为海洋网箱养殖的智能化管理提供了重要的技术支持。

关键词: 水产养殖网箱, 小目标检测, 自适应密集感知算法, 增强型快速空间金字塔池化, 混合注意力模块

Abstract:

[Objective] Detecting dense and small aquaculture net cages in complex backgrounds is difficult, the purpose of this study is to build a specialized dataset and design a targeted detection model that enhances recognition accuracy and robustness for practical aquaculture management. [Methods] A dataset of aquaculture net cages was constructed using high-resolution remote sensing imagery collected from seven representative farming regions (Australia, Canada, Chile, Croatia, Greece, China, and the Faroe Islands), and Cage-YOLO, a deep learning model based on YOLOv5, was proposed for detecting dense and small aquaculture net cages. First, an adaptive dense perception algorithm was introduced, which automatically selects and generates feature maps that reflect the high-density distribution of small aquaculture net cages. Second, an enhanced module based on spatial pyramid pooling fast was integrated to effectively reduce background noise interference and improve global feature extraction capabilities. Finally, a mixed attention block was incorporated to further enhance the model's perception of dense and small objects. [Results and Discussions] Experimental results showed that the proposed Cage-YOLO achieved improvements over the original YOLOv5 in terms of precision, recall, and mean average precision by 5.6, 21.8, and 17.4 percentage points, respectively. The model size was maintained at 16.9 MB, demonstrating both strong performance and deployment advantages. [Conclusions] This study provides a new approach for dense and small object detection and offers technical support for the intelligent management of marine cage aquaculture.

Key words: aquaculture net cage, small object detection, adaptive dense perception algorithm, enhancement spatial pyramid pooling, mix attention block

中图分类号:

S967.3

张文博, 江一珏, 宋巍, 贺琪, 张文博. CAGE-YOLO：基于遥感影像的水产养殖网箱高密度小目标检测模型[J]. 智慧农业(中英文), 2026, 8(2): 98-117.

ZHANG Wenbo, JIANG Yijue, SONG Wei, HE Qi, ZHANG Wenbo. CAGE-YOLO: A Dense Small Object Detection Model for Aquaculture Net Cages Based on Remote Sensing Images[J]. Smart Agriculture, 2026, 8(2): 98-117.

图/表 18

参考文献 40

[1]	DRACH A, TSUKROV I, DECEW J, et al. Engineering procedures for design and analysis of submersible fish cages with copper netting for exposed marine environment[J]. Aquacultural Engineering, 2016, 70: 1-14.
[2]	WEN L, CHENG Y, FANG Y, et al. A comprehensive survey of oriented object detection in remote sensing images[J]. Expert Systems with Applications, 2023, 224: 119960.
[3]	CHANDRA M A, BEDI S S. Survey on SVM and their application in imageclassification[J]. International Journal of Information Technology, 2021, 13(5): 1-11.
[4]	GIRSHICK R, DONAHUE J, DARRELL T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]// 2014 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway, New Jersey, USA: IEEE, 2014: 580-587.
[5]	GIRSHICK R. Fast R-CNN[C]// 2015 IEEE International Conference on Computer Vision (ICCV). Piscataway, New Jersey, USA: IEEE, 2015: 1440-1448.
[6]	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.
[7]	LIU W, ANGUELOV D, ERHAN D, et al. SSD: Single shot MultiBox detector[C]// Computer Vision-ECCV 2016. Cham, Germany: Springer, 2016: 21-37.
[8]	REDMON J, DIVVALA S, GIRSHICK R, et al. You only look once: Unified, real-time object detection[C]// 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway, New Jersey, USA: IEEE, 2016: 779-788.
[9]	REDMON J, FARHADI A. YOLO9000: Better, faster, stronger[C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway, New Jersey, USA: IEEE, 2017: 6517-6525.
[10]	REDMON J, FARHADI A. YOLOv3: An incremental improvement[EB/OL]. arXiv: 1804.02767, 2018.
[11]	BOCHKOVSKIY A, WANG C Y, LIAO H M. YOLOv4: Optimal speed and accuracy of object detection[EB/OL]. arXiv: 2004.10934, 2020.
[12]	GE Z, LIU S T, WANG F, et al. YOLOX: Exceeding YOLO series in 2021[EB/OL]. arXiv: 2107.08430, 2021.
[13]	MIAO L Z, LI N, ZHOU M L, et al. CBAM-Yolov5: Improved Yolov5 based on attention model for infrared ship detection[C]// International Conference on Computer Graphics, Artificial Intelligence, and Data Processing (ICCAID 2021). Burlingame, California, USA: SPIE, 2022: 33.
[14]	WANG M, YANG W Z, WANG L J, et al. FE-YOLOv5: Feature enhancement network based on YOLOv5 for small object detection[J]. Journal of Visual Communication and Image Representation, 2023, 90: 103752.
[15]	KIM M, JEONG J, KIM S. ECAP-YOLO: Efficient channel attention pyramid YOLO for small object detection in aerial image[J]. Remote Sensing, 2021, 13(23): 4851.
[16]	LUO X D, WU Y Q, WANG F Y. Target detection method of UAV aerial imagery based on improved YOLOv5[J]. Remote Sensing, 2022, 14(19): 5063.
[17]	SZEGEDY C, LIU W, JIA Y Q, et al. Going deeper with convolutions[C]// 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway, New Jersey, USA: IEEE, 2015: 1-9.
[18]	TIAN Z, SHEN C H, CHEN H, et al. FCOS: Fully convolutional one-stage object detection[C]// 2019 IEEE/CVF International Conference on Computer Vision (ICCV). Piscataway, New Jersey, USA: IEEE, 2020: 9626-9635.
[19]	KONG T, SUN F C, LIU H P, et al. FoveaBox: Beyound anchor-based object detection[J]. IEEE Transactions on Image Processing, 2020, 29: 7389-7398.
[20]	LAW H, DENG J. CornerNet: Detecting objects as paired keypoints[C]// Computer Vision-ECCV 2018. Cham, Germany: Springer, 2018: 765-781.
[21]	LU X, LI B Y, YUE Y X, et al. Grid R-CNN[C]// 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway, New Jersey, USA: IEEE, 2020: 7355-7364.
[22]	YANG Z, LIU S H, HU H, et al. RepPoints: Point set representation for object detection[C]// 2019 IEEE/CVF International Conference on Computer Vision (ICCV). Piscataway, New Jersey, USA: IEEE, 2020: 9656-9665.
[23]	CARION N, MASSA F, SYNNAEVE G, et al. End-to-end object detection with transformers[C]// Computer Vision-ECCV 2020. Cham, Germany: Springer, 2020: 213-229.
[24]	ZHU X Z, SU W J, LU L W, et al. Deformable DETR: Deformable transformers for end-to-end object detection[EB/OL]. arXiv: 2010.04159, 2020.
[25]	SUN P Z, ZHANG R F, JIANG Y, et al. Sparse R-CNN: End-to-end object detection with learnable proposals[C]// 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway, New Jersey, USA: IEEE, 2021: 14449-14458.
[26]	BAI Y C, ZHANG Y Q, DING M L, et al. SOD-MTGAN: Small object detection via multi-task generative adversarial network[C]// Computer Vision-ECCV 2018. New York, USA: ACM, 2018: 210-226.
[27]	BAI Y C, ZHANG Y Q, DING M L, et al. Finding tiny faces in the wild with generative adversarial network[C]// 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway, New Jersey, USA: IEEE, 2018: 21-30.
[28]	LI J N, LIANG X D, WEI Y C, et al. Perceptual generative adversarial networks for small object detection[C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway, New Jersey, USA: IEEE, 2017: 1951-1959.
[29]	LI C L, YANG T, ZHU S J, et al. Density map guided object detection in aerial images[C]// 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW). Piscataway, New Jersey, USA: IEEE, 2020: 737-746.
[30]	HUANG S Q, LIU Q. Addressing scale imbalance for small object detection with dense detector[J]. Neurocomputing, 2022, 473: 68-78.
[31]	ZHANG X D, IZQUIERDO E, CHANDRAMOULI K. Dense and small object detection in UAV vision based on cascade network[C]// 2019 IEEE/CVF International Conference on Computer Vision Workshop (ICCVW). Piscataway, New Jersey, USA: IEEE, 2020: 118-126.
[32]	HE K M, ZHANG X Y, REN S Q, et al. Spatial pyramid pooling in deep convolutional networks for visual recognition[C]// Computer Vision – ECCV 2014. Cham, Germany: Springer, 2014: 346-361.
[33]	LIU S, QI L, QIN H F, et al. Path aggregation network for instance segmentation[C]// 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway, New Jersey, USA: IEEE, 2018: 8759-8768.
[34]	LIN T Y, DOLLÁR P, GIRSHICK R, et al. Feature pyramid networks for object detection[C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway, New Jersey, USA: IEEE, 2017: 936-944.
[35]	ZHENG Z H, WANG P, LIU W, et al. Distance-IoU loss: Faster and better learning for bounding box regression[J]. Proceedings of the AAAI Conference on Artificial Intelligence, 2020, 34(7): 12993-13000.
[36]	LIU K, HUANG J, LI X L. Eagle-eye-inspired attention for object detection in remote sensing[J]. Remote Sensing, 2022, 14(7): 1743.
[37]	FANG Q Y, WANG Z K. Cross-modality attentive feature fusion for object detection in multispectral remote sensing imagery[J]. Pattern Recognition, 2022, 130: 108786.
[38]	WANG Z M, WANG J S, YANG K, et al. Semantic segmentation of high-resolution remote sensing images based on a class feature attention mechanism fused with Deeplabv3+[J]. Computers & Geosciences, 2022, 158: 104969.
[39]	HOU Q B, ZHOU D Q, FENG J S. Coordinate attention for efficient mobile network design[C]// 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway, New Jersey, USA: IEEE, 2021: 13708-13717.
[40]	WANG Q L, WU B G, ZHU P F, et al. ECA-net: Efficient channel attention for deep convolutional neural networks[C]// 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Piscataway, New Jersey, USA: IEEE, 2020: 11531-11539.

Method	P/%	R/%	mAP@0.5/%	mAP@0.5:0.95/%	Model size/MB
YOLOv5	89.9	71.5	77.8	41.5	14.4
YOLOv6s	84.8	68.7	71.5	35.3	38.1
YOLOv7	90.8	81.3	87.9	47.4	72.0
YOLOv8s	91.5	77.7	84.7	47.6	22.5
YOLOv9s	84.8	73.7	77.7	41.0	26.7
YOLOv10s	88.6	77.8	82.7	45.9	28.8
YOLOv11s	90.0	78.1	84.4	48.2	37.6
YOLOv12s	87.7	79.4	83.2	47.2	19.1
Faster R-CNN	69.3	77.0	69.3	35.5	108.0
DERT	89.3	77.1	83.2	45.3	158.0
Cage-YOLO	95.5	93.3	95.2	54.9	16.9

Baseline	ADPA	ESPPF	MAB	P/%	R/%	mAP@0.5/%	mAP@0.5:0.95/%
√	—	—	—	89.9	71.5	77.8	41.5
√	√	—	—	90.6	83.9	87.1	48.2
√	—	√	—	95.3	88.2	92.9	54.9
√	—	—	√	89.6	72.8	78.2	40.6
√	√	—	√	94.4	90.5	93.2	54.6
√	—	√	√	95.9	88.7	91.9	54.1
√	√	√	—	93.9	85.3	90.7	51.0
√	√	√	√	95.5	93.3	95.2	54.9

Baseline	ADPA	ESPPF	MAB	circle mAP@0.5/%	rectangular mAP@0.5/%	circle mAP@0.5:0.95/%	rectangular mAP@0.5:0.95/%
√	—	—	—	76.4	95.4	40.8	53.2
√	√	—	—	83.1	96.2	45.1	54.0
√	—	√	—	92.2	96.3	55.0	52.6
√	—	—	√	77.2	95.4	40.6	52.2
√	√	—	√	90.7	95.3	51.5	53.6
√	—	√	√	90.8	96.4	54.8	52.1
√	√	√	—	90.6	93.7	53.4	50.2
√	√	√	√	93.1	95.5	55.0	53.4

YOLOv5+ESPPF+ADPA	CA	ECA	CBAM	MAB	P/%	R/%	mAP@0.5/%	mAP@0.5:0.95/%
√	√	—	—	—	96.1	88.0	91.5	53.0
√	—	√	—	—	96.4	88.6	92.0	53.9
√	—	—	√	—	95.5	89.6	92.4	54.0
√	—	—	—	√	95.5	93.3	95.2	54.9

Method	P/%	R/%	mAP@0.5/%	mAP@0.5:0.95/%
Cage-YOLO	95.5	93.3	95.2	54.9
Cage-YOLOv8s	87.8	78.5	82.7	46.7
Cage-YOLOv10s	87.8	77.4	81.1	46.5
Cage-YOLOv11s	89.5	77.9	83.9	47.0
Cage-YOLOv12s	87.9	72.3	78.9	43.8