基于边缘计算和改进MobileNet v3的奶牛反刍行为实时监测方法

doi:10.12133/j.smartag.SA202405023

Smart Agriculture ›› 2024, Vol. 6 ›› Issue (4): 29-41.doi: 10.12133/j.smartag.SA202405023

• 专题--智慧畜牧技术创新与可持续发展 • 上一篇下一篇

基于边缘计算和改进MobileNet v3的奶牛反刍行为实时监测方法

张宇¹, 李相廷¹, 孙雅琳², 薛爱迪¹^,³, 张翼¹, 姜海龙¹, 沈维政¹()

^1. 东北农业大学电气与信息学院，黑龙江哈尔滨 150030，中国
^2. 哈尔滨航天恒星数据系统科技有限公司，黑龙江哈尔滨 150030，中国
^3. 哈尔滨电机厂有限责任公司，黑龙江哈尔滨 150030，中国

收稿日期:2024-05-30 出版日期:2024-07-30
基金项目:
国家重点研发计划项目(2023YFD2000700); 财政部和农业农村部：国家现代农业产业技术体系资助(CARS36)
作者简介:
张宇，E-mail: zhangyu1900@neau.edu.cn
通信作者:
沈维政，E-mail: wzshen@neau.edu.cn

Real-Time Monitoring Method for Cow Rumination Behavior Based on Edge Computing and Improved MobileNet v3

ZHANG Yu¹, LI Xiangting¹, SUN Yalin², XUE Aidi¹^,³, ZHANG Yi¹, JIANG Hailong¹, SHEN Weizheng¹()

^1. College of Electrical and Information, Northeast Agricultural University, Harbin 150030, China
^2. Harbin Aerospace Stellar Data System Technology Co. , Ltd, Harbin 150030, China
^3. Harbin Electric Machinery Company Co. , Ltd. , Harbin 150030, China

Received:2024-05-30 Online:2024-07-30
Foundation items:The National Key Research and Development Program of China(2023YFD2000700); Supported by The Earmarked Fund for CARS36(CARS36)
About author:
ZHANG Yu, research interests are agricultural AI and AIoT. E-mail: zhangyu1900@neau.edu.cn
Corresponding author:
SHEN Weizheng, PhD, Professor, research interest is smart livestock. E-mail: wzshen@neau.edu.cn

摘要/Abstract

摘要：

[目的/意义] 随着奶牛养殖业向规模化、精准化和信息化养殖迅速发展，对奶牛健康的监测和管理需求也日益增加。实时监测奶牛的反刍行为对于第一时间获取奶牛健康的相关信息以及预测奶牛疾病具有至关重要的意义。目前，针对奶牛反刍行为的监测已经提出了多种策略，包括基于视频监控、声音识别、传感器监测等方法，但是这些方法普遍存在实时性不足的问题。为了减轻数据传输的数量与云端计算量，实现对奶牛反刍行为的实时监测，基于边缘计算的思想提出了一种实时对奶牛反刍行为进行监测的方法。 [方法] 使用自主设计的边缘设备实时地采集并处理奶牛的六轴加速度信号，基于六轴数据提出了基于联邦式与拆分式边缘智能这两种不同的策略对奶牛反刍行为实时识别方法展开研究。在基于联邦式边缘智能的奶牛反刍行为实时识别方法研究中，通过协同注意力机制改进MobileNet v3网络提出了CA-MobileNet v3网络，进而利用CA-MobileNet v3网络和FedAvg模型聚合算法，设计了联邦式边缘智能模型。在基于拆分式边缘智能的奶牛反刍行为实时识别方法研究中，利用融合协同注意力机制的MobileNet v3网络和Bi-LSTM网络，设计了基于MobileNet-LSTM的拆分式边缘智能模型。 [结果和讨论] 对比了MobileNet v3、CA-MobileNet、联邦式边缘智能模型，以及拆分式边缘智能模型的识别准确率，其中基于CA-MobileNet v3的联邦式边缘智能模型的平均查准率、召回率、 F₁-Score、特异性以及准确率分别达到97.1%、97.9%、97.5%、98.3%和98.2%，达到了最佳识别效果。 [结论] 本研究为奶牛反刍行为的监测提供了一种实时有效的方法，所提出的方法可以在实际应用中使用。

关键词: 奶牛反刍行为, 实时监测, 边缘计算, 改进MobileNet v3, 边缘智能模型, Bi-LSTM

Abstract:

[Objective] Real-time monitoring of cow ruminant behavior is of paramount importance for promptly obtaining relevant information about cow health and predicting cow diseases. Currently, various strategies have been proposed for monitoring cow ruminant behavior, including video surveillance, sound recognition, and sensor monitoring methods. However, the application of edge device gives rise to the issue of inadequate real-time performance. To reduce the volume of data transmission and cloud computing workload while achieving real-time monitoring of dairy cow rumination behavior, a real-time monitoring method was proposed for cow ruminant behavior based on edge computing. [Methods] Autonomously designed edge devices were utilized to collect and process six-axis acceleration signals from cows in real-time. Based on these six-axis data, two distinct strategies, federated edge intelligence and split edge intelligence, were investigated for the real-time recognition of cow ruminant behavior. Focused on the real-time recognition method for cow ruminant behavior leveraging federated edge intelligence, the CA-MobileNet v3 network was proposed by enhancing the MobileNet v3 network with a collaborative attention mechanism. Additionally, a federated edge intelligence model was designed utilizing the CA-MobileNet v3 network and the FedAvg federated aggregation algorithm. In the study on split edge intelligence, a split edge intelligence model named MobileNet-LSTM was designed by integrating the MobileNet v3 network with a fusion collaborative attention mechanism and the Bi-LSTM network. [Results and Discussions] Through comparative experiments with MobileNet v3 and MobileNet-LSTM, the federated edge intelligence model based on CA-MobileNet v3 achieved an average Precision rate, Recall rate, F₁-Score, Specificity, and Accuracy of 97.1%, 97.9%, 97.5%, 98.3%, and 98.2%, respectively, yielding the best recognition performance. [Conclusions] It is provided a real-time and effective method for monitoring cow ruminant behavior, and the proposed federated edge intelligence model can be applied in practical settings.

Key words: cow rumination behavior, real-time monitoring, edge computing, improved MobileNet v3, edge intelligence model, Bi-LSTM

张宇, 李相廷, 孙雅琳, 薛爱迪, 张翼, 姜海龙, 沈维政. 基于边缘计算和改进MobileNet v3的奶牛反刍行为实时监测方法[J]. 智慧农业(中英文), 2024, 6(4): 29-41.

ZHANG Yu, LI Xiangting, SUN Yalin, XUE Aidi, ZHANG Yi, JIANG Hailong, SHEN Weizheng. Real-Time Monitoring Method for Cow Rumination Behavior Based on Edge Computing and Improved MobileNet v3[J]. Smart Agriculture, 2024, 6(4): 29-41.

图/表 16

参考文献 27

1	BRAUN U, TSCHONER T, HÄSSIG M. Evaluation of eating and rumination behaviour using a noseband pressure sensor in cows during the peripartum period[J]. BMC veterinary research, 2014, 10: ID195.
2	HE M N. Research on key technologies for digital management of cattle[D]. Jinan: Shandong University.
3	JI N, YIN Y L, SHEN W Z, et al. Pig sound analysis: A measure of welfare[J]. Smart Agriculture, 2022, 4( 2): 19- 35.
4	TANI Y, YOKOTA Y, YAYOTA M, et al. Automatic recognition and classification of cattle chewing activity by an acoustic monitoring method with a single-axis acceleration sensor[J]. Computers and electronics in agriculture, 2013, 92: 54- 65.
5	VÁZQUEZ DIOSDADO J A, BARKER Z E, HODGES H R, et al. Classification of behaviour in housed dairy cows using an accelerometer-based activity monitoring system[J]. Animal biotelemetry, 2015, 3( 1): 15.
6	BENAISSA S, TUYTTENS F A M, PLETS D, et al. Classification of ingestive-related cow behaviours using RumiWatch halter and neck-mounted accelerometers[J]. Applied animal behaviour science, 2019, 211: 9- 16.
7	SHEN W Z, CHENG F, ZHANG Y, et al. Automatic recognition of ingestive-related behaviors of dairy cows based on triaxial acceleration[J]. Information processing in agriculture, 2020, 7( 3): 427- 443.
8	HOU S. Research on real-time recognition of cattle ruminating behavior based on activity data and neural networks[D]. Hohhot: Inner Mongolia University, 2021.
9	WANG J H, HAN Y X. Cognitive radio sensor networks clustering routing algorithm for crop phenotypic information edge computing collection[J]. Smart agriculture, 2020, 2( 2): 28- 47.
10	CHIANG M, ZHANG T. Fog and IoT: An overview of research opportunities[J]. IEEE Internet of Things journal, 2016, 3( 6): 854- 864.
11	SHI W S, CAO J, ZHANG Q, et al. Edge computing: Vision and challenges[J]. IEEE Internet of Things journal, 2016, 3( 5): 637- 646.
12	ZHU Y. Network public opinion prediction and control based on edge computing and artificial intelligence new paradigm[J]. Wireless communications and mobile computing, 2021, 2021: ID 5566647.
13	BONAWITZ K, EICHNER H, GRIESKAMP W, et al. Towards federated learning at scale: system design[J]. Proceedings of the 2019 conference on systems and machine learning, 2019, 1: 374- 388.
14	WANG T, SUN B, ZHANG Y L, et al. A method for device assessment and federated learning importance aggregation based on edge intelligence, system, device, and readable storage medium[P]. CN 112181666A, 2021.
15	DUAN Q, HU S, DENG R, et al. Combined federated and split learning in edge computing for ubiquitous intelligence in Internet of Things: State-of-the-art and future directions[J]. Sensors (basel, Switzerland), 2022, 22( 16): ID 5983.
16	FATEMI MOGHADDAM F, MOGHADDAM S G, ROUZBEH S, et al. A scalable and efficient user authentication scheme for cloud computing environments[C]// 2014 IEEE REGION 10 SYMPOSIUM. Piscataway, New Jersey, USA: IEEE, 2014: 508- 513.
17	HUANG C, KE Y, HUA X, et al. Current applications and prospects of edge computing in smart agriculture[J]. Transactions of the Chinese society of agricultural engineering, 2022, 38( 16): 224- 234.
18	CHEN G, WANG Z X, WANG J. An AIOT-based comprehensive application system for smart agriculture[J]. Automation applications, 2022( 7): 48- 50.
19	LI Y H. Design and implementation of an optimization system for real-time data processing applications in edge computing mode[M]. Nanjing: Southeast University, 2019.
20	BU F Y, WANG X. A smart agriculture IoT system based on deep reinforcement learning[J]. Future generation computer systems, 2019, 99: 500- 507.
21	SHEN W Z, SUN Y L, ZHANG Y, et al. Automatic recognition method of cow ruminating behaviour based on edge computing[J]. Computers and electronics in agriculture, 2021, 191: ID 106495.
22	WEN K, YU K G, LI Y B, et al. A new quaternion Kalman filter based foot-mounted IMU and UWB tightly-coupled method for indoor pedestrian navigation[J]. IEEE transactions on vehicular technology, 2020, 69( 4): 4340- 4352.
23	DISSANAYAKE G, SUKKARIEH S, NEBOT E, et al. The aiding of a low-cost strapdown inertial measurement unit using vehicle model constraints for land vehicle applications[J]. IEEE transactions on robotics and automation, 2001, 17( 5): 731- 747.
24	LI Y G, ZHANG S G, ZHU B, et al. Accurate human activity recognition with multi-task learning[J]. CCF transactions on pervasive computing and interaction, 2020, 2( 4): 288- 298.
25	MAHAN B MC, MOORE E, RAMAGE D, et al. Communication-efficient learning of deep networks from decentralized data[C]// International Conference on Artificial Intelligence and Statistics. Fort Lauderdale, Florida, USA: AISTATS, 2017: 1273- 1282.
26	SHI B, BAI X, YAO C. An end-to-end trainable neural network for image-based sequence recognition and its application to scene text recognition[J]. IEEE transactions on pattern analysis and machine intelligence, 2017, 39( 11): 2298- 2304.
27	XIONG F, CHEN T, BIAN B C, et al. Chip surface character recognition based on convolutional recurrent neural network[J]. Journal of Zhejiang university, 2023, 57( 5): 1- 9.

Feature types	Feature quantity	Feature description
Minimum value	9	Triaxial acceleration， triaxial angular velocity， SMV， attitude Angle
The first quartile	9	Triaxial acceleration， triaxial angular velocity， SMV， attitude Angle
Median value	9	Triaxial acceleration， triaxial angular velocity， SMV， attitude Angle
Third quartile	9	Triaxial acceleration， triaxial angular velocity， SMV， attitude Angle
Maximum value	9	Triaxial acceleration， triaxial angular velocity， SMV， attitude Angle
Mean value	9	Triaxial acceleration， triaxial angular velocity， SMV， attitude Angle
Root mean square	9	Triaxial acceleration， triaxial angular velocity， SMV， attitude Angle
Standard deviation	9	Triaxial acceleration， triaxial angular velocity， SMV， attitude Angle
Mean absolute deviation	9	Triaxial acceleration， triaxial angular velocity， SMV， attitude Angle
Coefficient of correlation	6	Triaxial acceleration， triaxial angular velocity
Spectral energy	3	SMV， attitude Angle
Frequency domain entropy	3	SMV， attitude Angle
Direct current component	3	SMV， attitude Angle

Input	Operation	Exp size	Output	CA	NL	Step length
96×96×1	conv2d，3×3	—	16	—	HS	2
48×48×16	bneck，3×3	16	16	√	RE	2
24×24×16	bneck，3×3	72	24	—	RE	2
12×12×24	bneck，3×3	88	24	—	RE	1
12×12×24	bneck，3×3	96	40	√	HS	2
6×6×40	bneck，5×5	240	40	√	HS	1
6×6×40	bneck，5×5	240	40	√	HS	1
6×6×40	bneck，5×5	120	48	√	HS	1
6×6×48	bneck，5×5	144	48	√	HS	1
6×6×48	bneck，5×5	288	96	√	HS	2
3×3×96	bneck，5×5	576	96	√	HS	1
3×3×96	bneck，5×5	576	96	√	HS	1
3×3×96	conv2d，1×1	—	576	√	HS	1
3×3×576	pool，3×3	—	—	—	—	1
1×1×576	conv2d 1×1，NBN	—	1 024	—	HS	1
1×1×1 024	conv2d 1×1，NBN	—	1	—	—	1

Performance metrics	MoblieNet v3	CA-MoblieNet v3
Precision/%	90.5	95.8
Recall/%	91.6	93.6
F ₁-Score/%	91.1	94.7
Specificity/%	94.5	97.7
Accuracy/%	93.5	96.2

Performance metrics	non-federated learning model	federated learning model
Precision/%	95.0	97.1
Recall/%	93.3	97.9
F ₁-Score/%	94.1	97.5
Specificity/%	97.2	98.3
Accuracy/%	95.8	98.2

Performance metrics/%	Federated edge intelligence model	Split edge intelligence model
Precision	97.1	95.8
Recall	97.9	93.7
F ₁-Score	97.5	94.8
Specificity	98.3	97.7
Accuracy	98.2	96.2

基于边缘计算和改进MobileNet v3的奶牛反刍行为实时监测方法

Real-Time Monitoring Method for Cow Rumination Behavior Based on Edge Computing and Improved MobileNet v3

在线阅读

知网下载

本地下载

可视化

摘要/Abstract

引用本文

使用本文

图/表 16

参考文献 27

相关文章 2

编辑推荐

Metrics

本文评价

[1]	张荣华, 白雪, 樊江川. 复杂场景下害虫目标检测算法：YOLOv8-Extend[J]. 智慧农业(中英文), 2024, 6(2): 49-61.
[2]	夏雪, 柴秀娟, 张凝, 周硕, 孙琦鑫, 孙坦. 用于边缘计算设备的果树挂果量轻量化估测模型[J]. 智慧农业(中英文), 2023, 5(2): 1-12.