改进YOLOv5的活性污泥微生物镜检方法

张宇航; 苑明哲; 王文洪; 张佳; 肖金超; 曹飞道

doi:10.13205/j.hjgc.202506019

改进YOLOv5的活性污泥微生物镜检方法

doi: 10.13205/j.hjgc.202506019

张宇航^1,2,3,
苑明哲^2,3, ,,
王文洪³,
张佳¹,
肖金超^2,3,
曹飞道³

1. 沈阳工业大学信息科学与工程学院, 沈阳 110870;
2. 中国科学院沈阳自动化研究所, 沈阳 110016;
3. 广州工业智能研究院, 广州 511458

基金项目:

国家自然科学基金面上项目（62273332）；国家自然科学基金面上项目（62273337）；中科院科技服务网络计划（STS）-东莞专项（20211600200072）

详细信息

作者简介:
张宇航（2000—），硕士研究生，主要研究方向为机器视觉、深度学习、智能制造。zhangyuhang@gz.sia.cn

通讯作者:
苑明哲（1971—），博士，研究员，主要研究方向为工业物联网、分布式控制系统和工业过程控制领域。mzyuan@sia.cn

计量
- 文章访问数: 58
- HTML全文浏览量: 14
- PDF下载量: 1
- 被引次数: 0
出版历程
- 收稿日期: 2024-09-28
- 录用日期: 2024-12-10
- 修回日期: 2024-11-08

An improved YOLOv5-based microscopic examination method for activated sludge microorganisms

1. School of Information Science and Engineering, Shenyang University of Technology, Shenyang 110870, China;
2. Shenyang Institute of Automation, Chinese Academy of Sciences, Shenyang 110016, China;
3. Guangzhou Industrial Intelligence Research Institute, Guangzhou 511458, China

摘要

摘要: 针对现有活性污泥微生物检测算法精度低，漏检率较高的问题，提出了一种改进YOLOv5的微生物检测方法。利用K-Means++算法生成最适合本数据集的锚框组，在YOLOv5主干网络中用C3GC模块代替原有C3模块加强特征信息提取，颈部网络的特征金字塔融合坐标注意力机制与全局到空间聚合模块加强特征信息的融合。基于实际采集的活性污泥微生物数据，通过数据增强方法构建了训练和测试数据集。改进算法在测试数据集上召回率达到97.4%，平均精准度达到99.2%，模型大小仅为23.5 MB，与原始YOLOv5模型和其他主流检测模型及变种相比，准确率与回归率均取得了较大提升，并能够满足在移动端的快速部署，证明了改进算法的有效性。
- 污水处理 /
- 微生物检测 /
- YOLOv5 /
- 特征金字塔 /
- 坐标注意力机制 /
- 全局到空间聚合模块
Abstract: To address the research shortages， as well as the low accuracy and high missed detection rate of existing activated sludge microbial detection algorithms， an improved YOLOv5-based microbial detection method was proposed. The K-Means++ algorithm was used to generate the most suitable anchor frame group for this dataset. The original C3 module in the YOLOv5 backbone network was replaced with the C3GC module to enhance feature extraction. In addition， the feature pyramid fusion coordinate attention mechanism in the neck network and the global-to-spatial aggregation module were employed to strengthen the fusion of feature information. Based on the actual collected microbial data of activated sludge， the training and test datasets were constructed using a data augmentation method. The improved algorithm achieved a recall rate of 97.4%， an average accuracy of 99.2%， and a model size of only 23.5MB on the test dataset， which demonstrated a certain improvement in accuracy and regression rate compared to the original YOLOv5 model and other mainstream detection models and variants， while meeting the requirements for rapid deployment on mobile terminals. The experimental data showed that the improved model sacrificed a small amount of detection speed in exchange for a significant improvement in accuracy and regression rate， proving the effectiveness of the improvement. This study provides a valuable reference for the development of artificial intelligence in the field of environmental governance and serves as an example for the advancement of intelligent equipment.
- sewage treatment /
- microbial detection /
- YOLOv5 /
- characteristic pyramids /
- coordinate attention mechanism /
- global-to-spatial aggregation module

HTML全文

参考文献(19)

[1]	China Environmental Health and Safety Net. Statistical communiqué of emission sources in Guangzhou in 2022［EB/OL］. ehs.cn，2023-11-16. https://www.ehs.cn/article-22489-1.html 中国环境健康安全网. 广州市2022年排放源统计公报［EB/OL］. ehs.cn，2023-11-16. https://www.ehs.cn/article-22489-1.html
[2]	YUAN P S，ZHANG H R，CHEN X，et al. Comparison of two improved activated sludge methods in urban sewage treatment［J］. Environmental Engineering，2006，24（3）:34-35. 原培胜，张洪荣，陈现明，等. 两种改良型活性污泥法在城市污水处理中的比较［J］. 环境工程，2006，24（3）:34-35.
[3]	ZHOU C，YAN P，YUAN S，et al. Intelligent microscopic examination device and microscopic examination method for activated sludge microorganisms:CN201910251070.7［P］. 2024-08-14. 周仓，闫攀，袁帅，等. 一种用于活性污泥微生物的智能镜检装置及镜检方法:CN201910251070.7［P］. 2024-08-14.
[4]	Wu S，XU Y，ZHAO D N. Survey of Object Detection Based on Deep Convolutional Network［J］. 2018，31（4）:335-346. 吴帅，徐勇，赵东宁. 基于深度卷积网络的目标检测综述［J］. 模式识别与人工智能，2018，31（4）:335-346.
[5]	Girshick R. Fast R-CNN［J］. Computer Science，2015. DOI: 10.1109/ICCV.2015.169.
[6]	HE K，GKIOXARI G，PIOTR D，et al. Mask R-CNN［J］. IEEE Transactions on Pattern Analysis & Machine Intelligence，202，42（2）:386-397.
[7]	TERVEN J，CÓRDOVA-ESPARZA D M，ROMERO-GONZÁLEZ J A. A comprehensive review of yolo architectures in computer vision:From YOLOv1 to YOLOv8 and Yolo-Nas［J］. Machine Learning and Knowledge Extraction，2023，5（4）:1680-1716.
[8]	GUO G，ZHANG Z. Road damage detection algorithm for improved YOLOv5［J］. Scientific reports，2022，12（1）:15523.
[9]	ZHAO M，YU H，LI H Q，et al. Detection of fish stocks by fused with SKNet and YOLOv deep learning［J］. Journal of Dalian Fisheries University，2022，37（2）:312-319. 赵梦，于红，李海清，等. 融合SKNet与YOLOv5深度学习的养殖鱼群检测［J］. 大连海洋大学学报，2022，37（2）:312-319.
[10]	YANG Q S，LI W K，YANG X F，et al. Improved YOLOv5 method for detecting growth status of apple flowers［J］. Computer Engineering and Applications，2022，58（4）:237-246. 杨其晟，李文宽，杨晓峰，等. 改进YOLOv5的苹果花生长状态检测方法［J］. 计算机工程与应用，2022，58（4）:237-246.
[11]	SONG H B，WANG Y F，DUAN Y C，et al. Detection method of severe adhesive wheat grain based on YOLOv5-MDC Model［J］. Transactions of the Chinese Society for Agricultural Machinery，2022，53（4）:245-253. 宋怀波，王云飞，段援朝，等. 基于YOLOv5-MDC的重度粘连小麦籽粒检测方法［J］. 农业机械学报，2022（4）:245-253.
[12]	WANG J，SUN Z Y，GUO P，et al. Improved leukocyte detection algorithm of YOLOv5［J］. Computer Engineering and Applications，2022，58（4）:134-142. 王静，孙紫雲，郭苹，等. 改进YOLOv5的白细胞检测算法［J］. 计算机工程与应用，2022，58（4）:134-142.
[13]	LIN T Y，DOLLAR P，GIRSHICK R，et al. Feature pyramid networks for object detection［J］. IEEE Computer Society，2017. DOI: 10.1109/CVPR.2017.106.
[14]	LIU S，QI L，QIN H，et al. Path aggregation network for instance segmentation［J］. IEEE，2018. DOI: 10.1109/CVPR.2018.00913..
[15]	CAO Y，XU J，LIN S，et al. GCNet:Non-local networks meet squeeze-excitation networks and beyond［C］// 2019 IEEE/CVF International Conference on Computer Vision Workshop（ICCVW）. IEEE，2020. DOI: 10.1109/ICCVW.2019.00246.
[16]	HOU Q，ZHOU D，FENG J. Coordinate attention for efficient mobile network design［EB/OL］. arXiv，2021. DOI: 10.48550/arXiv.2103.02907.
[17]	TANG F，XU Z，HUANG Q，et al. DuAT:Dual-aggregation transformer network for medical image segmentation［C］// Chinese Conference on Pattern Recognition and Computer Vision（PRCV）. Singapore:Springer Nature Singapore，2023:343-356.
[18]	ARTHUR D，VASSILVITSKII S. K-means++:the advantages of careful seeding［C］// Proceedings of the Eighteenth Annual ACM-SIAM Symposium on Discrete Algorithms. New Orleans，SIAM，2007:1027-1035.
[19]	IKOTUN A M，EZUGWU A E，ABUALIGAH L，et al. K-means clustering algorithms:a comprehensive review，variants analysis，and advances in the era of big data［J］. Information Sciences，2023，622:178-210.