MOOnitor: An IoT based multi-sensory intelligent device for cattle activity monitoring

https://doi.org/10.1016/j.sna.2021.113271Get rights and content

Highlights

  • IoT enabled intelligent multi-sensory device for cattle activity monitoring is developed.

  • Activity classification is conducted using XGBoost and Random Forest classifier.

  • Temperature and speed data along with acceleration values classification accuracy.

  • Classification accuracy of ~97% obtained on use of XGBoost classifier.

Abstract

Continuous activity monitoring of dairy cattle is essential to acquire a comprehensive knowledge on health and well-being of the animals. In this research, we have reported the development and deployment of "MOOnitor", a neck-mounted intelligent IoT device for cattle monitoring. The device facilitates classification of salient activities of cattle through appropriately positioned sensors. MOOnitor is an integration of a temperature sensor, a global positioning system (GPS) module, and a 3-axis accelerometer in a lightweight enclosure, which is attached to a halter that allows transmission of data to an IoT server using a microcontroller and a cellular GSM module. After acquiring the necessary sensory information, the most significant features were strategically extracted for enhanced data interpretation. Thereafter, optimally tuned eXtreme Gradient Boosting (XGBoost) and Random Forests classifiers were implemented to classify activities like ‘standing’, ‘lying’, ‘standing and ruminating’, ‘lying and ruminating’, ‘walking’, and ‘walking and grazing’. The performances of the two classifiers towards identification of different cattle activities were compared in terms of accuracy. Furthermore, the importance of using a temperature sensor and a GPS module in addition to an accelerometer in cattle activity recognition could be justified. An overall classification accuracy as high as ~97% was achieved using the XGBoost based classifier. In addition, accuracy, precision, sensitivity and specificity for standing (0.98, 0.97, 0.97, 0.98), lying (0.97, 0.90, 1, 0.96), standing and ruminating (0.99, 1, 0.97, 1), lying and ruminating (0.99, 1, 0.83, 1), walking (1, 1, 1, 1), and walking and grazing (0.99, 1, 0.75, 1) shows the suitability of the proposed method in effective cattle activity monitoring. Since cattle activity states are indicative of various factors such as estrous and several diseases like mastitis, foot-and-mouth disease, etc, the MOOnitor may be used for early detection of these conditions in addition to general health monitoring.

Introduction

Livestock plays a vital role in the socio-economic growth of the developing countries [3]. Recent years are marked with a significant increase in cattle population thereby arising the need for effective cattle monitoring technology so that diseases in animals can be anticipated and immediate actions can be imposed [1]. Regular monitoring of cattle primarily involves critical observation of day-to-day activities of cattle to draw suitable inferences related to health and well-being of the animals. In this context, the existing literature claims that anomalies in cattle activities such as standing, lying, grazing, ruminating, and walking for each animal can help to detect (re)production, health, and welfare problems such as onset of estrous, mastitis, lameness, foot-and-mouth diseases (FMD), etc. These practises have received notable appreciation in recent years with special emphasis on use of non-invasive wearable sensors which are robust enough to fit in free grazing environment [10], [5].

Existing literature entailing activity recognition of cattle mostly focuses on use of wearable accelerometer sensors; the use of acoustic sensors, RFID location tags, and cameras coupled with radio frequency modules for data transmission to nearby nodes are also found. Accelerometer sensors are generally placed in the neck, ear or leg of the animal [2]. The cattle activities that are generally supervised include standing, ruminating/ grazing, walking, lying or a combination of these states [9]. An account of the existing studies related to cattle activity monitoring is presented in Table 1.

A detailed study of advances in cattle activity monitoring using non-invasive sensors, as presented in Table 1, infers the following observations, which need to be considered for enhanced acceptance of the technology in real-time cattle monitoring- (a) Leg mounted accelerometer based devices are claimed to be less effective in classification of feeding and standing behaviour [6], as a result of which researchers have used combination of leg and collar mounted units to achieve better classification accuracy [11]. However, placement of sensors at two different locations is equally complex and challenging for continuous monitoring of animals. A neck mounted device that adequately classifies the aforementioned activities is hence desired for ease of access; (b) Research in veterinary science reveals that the body temperature of cows increases significantly during active hours irrespective of the duration [4]. For instance, a standing and ruminating cow exhibits a slightly higher temperature than a lying or resting one. However, current studies have not mainly focussed on this factor. Inclusion of a temperature sensor that would continually log a cow’s body temperature might lead to more accurate activity monitoring; (c) Majority of the reported research focuses on activity classification within farm or barn environment [12,[26], [12], [15]. However, in developing countries cattle are mostly found to move around freely in the outdoors during the day time which calls for a need of a sensory technology that would meet out-of-farm consequences. This is where the temperature and walking speed of the cattle become the dominant factors for activity analysis. Hence, installation of a Global Positioning System (GPS) module along with a temperature sensor would be a prospective solution; (d) The reported studies deal with the data collection and interpretation of cattle activity wherein the data is either manually offloaded [13] or transmitted via radio frequency modules [11]. This becomes impractical for walking and grazing cattle since data transmission is limited by the effective range of radio frequency module. A SIM based module which can provide cellular internet connectivity can alleviate this issue.

Considering the aforementioned findings, in this research we have presented “MOOnitor”- a device housing a temperature sensor, a 3-axis accelerometer sensor, and a GPS module measuring the cattle body temperature, acceleration due to movements, and walking speed respectively to effectuate seamless monitoring of various cattle activities in real-time scenario both in and off the farm. A group of healthy, lactating crossbred cows from three different barns was selected for experimental purpose. The MOOnitor was installed on a halter and fastened to the neck of each cow for data acquisition. An onboard microcontroller captured the necessary data and transmitted the information via a SIM based GSM module to a remote server. The acquired data was then subjected to feature extraction, following which, intelligent algorithms were deployed to draw suitable inferences.

Bagging and boosting algorithms are the two popular ensemble methods wherein multiple learning algorithms are used to train models with same dataset to achieve a better prediction accuracy, avoid overfitting, and effectively handle bias-variance trade-off. The RFs that comprises many decision trees are trained through bagging and the predictions are made based on the average of the output of the decision trees. This enhances the prediction accuracy and consequently, the limitations of decision trees are alleviated. On the other hand, XGBoost algorithm deals with gradient boosted decision trees for enhanced performance. Considering several factors such as regularization, apprehension of missing value, flexibility, etc. leading to speedy and accurate predictions, XGBoost based classification and regression algorithms are gaining importance in recent times. Here, the feature-extracted dataset was trained using optimally tuned XGBoost and RF classifier and the results of both were compared. Although the bagging and boosting algorithms are popular in related literature [6], [15], very few researchers have implemented XGBoost in sensor-driven cattle monitoring. For instance, [8] studied the feasibility of an automated detection system in classifying sick and healthy cattle using a motion sensor and necessary on-farm health recordings. Results of the research indicated a decent F1-measure of 81% and much lesser computational time allowing implementation of the methodology in online and/or cloud platforms [8]. However, classification of cattle activity using XGBoost is not attempted till date to the best of our knowledge. Motivated by the aforesaid findings and considering the exceptional performance of XGBoost based classification in other data-driven applications reported in recent literature [7], the authors implemented XGBoost in this research for classification of cattle activities. Further, benefits of using multi-sensory approach over accelerometery have been discussed to reach at interesting inferences.

Section snippets

Materials and methods

The MOOnitor is an assembly of a temperature sensor, a GPS module, and a 3-axis accelerometer on a microcontroller platform used for the measurement of associated physical quantities pertaining to cattle activities. The sensory information was transmitted to an IoT server using a GSM module for further interpretation. Intelligent algorithms were implemented thereafter to classify different activity states. The methodology is elaborated in the following sections. A block diagram representing the

Results and discussions

Observations related to validation of sensors, profiles for sensory information pertaining to various cattle activity and activity classification using machine intelligence are discussed in the following subsections.

Conclusions

In this research we have demonstrated the construction, working and deployment of a neck mounted intelligent IoT aided device called ‘MOOnitor’ for cattle activity monitoring inside the barn as well as in the pasture. The device is capable of recording acceleration, temperature and walking speed information of cattle and transmitting the data directly to a server over SIM based GSM module. Thereafter, implementation of intelligent algorithm with the acquired information would facilitate an

Author Statement

Debeshi Dutta (DD) conducted the hardware assembly and software integration of MOOnitor and worked for the manuscript preparation, Dwipjyoti Natta (DN) conducted field trials and field experiments, Dr. Soumen Mandal (SM) coordinated the work, was involved in IoT integration and manuscript preparation, Dr. Nilotpal Ghosh (NG) coordinated the work and was involved in field testing and trials of the MOOnitor device.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

The work was funded under "Promoting Innovations in Individuals, Start-ups and MSMEs (PRISM) Scheme" of "Department of Scientific and Industrial Research" (DSIR), Govt. of India, Grant No: DSIR/PRISM/70/2018 titled: “IoT based affordable cattle monitoring system for empowerment of Indian farmers”.

Debeshi Dutta completed B.Tech in Electronics and Communication Engineering from MAKAUT (formely WBUT) India and M.Tech in Biomedical Engineering from National Institute of Technology, Rourkela, India. She is currently pursuing Ph.D at AcSIR-CSIR CMERI, Durgapur, WB, India. She was a Visiting Scientist at Neuroimaging and Neuro informatics Laboratory at Kyushu University, Japan. She is an Innovator under PRISM scheme of DSIR, Govt of India and is working towards setting up of start-up using new

References (29)

Cited by (12)

View all citing articles on Scopus

Debeshi Dutta completed B.Tech in Electronics and Communication Engineering from MAKAUT (formely WBUT) India and M.Tech in Biomedical Engineering from National Institute of Technology, Rourkela, India. She is currently pursuing Ph.D at AcSIR-CSIR CMERI, Durgapur, WB, India. She was a Visiting Scientist at Neuroimaging and Neuro informatics Laboratory at Kyushu University, Japan. She is an Innovator under PRISM scheme of DSIR, Govt of India and is working towards setting up of start-up using new technologies and product development. Her research interests include wearable electronics, soft-robotics, rehabilitation engineering and machine learning.

Dwipjyoti Natta completed his graduation from College of Veterinary Science and Animal Husbandry, Tripura, India and Masters degree (MVSc) from West Bengal University of Animal and Fishery Sciences, Kolkata.

Dr. Soumen Mandal is a Senior Scientist at The Central Mechanical Engineering Research Institute, Durgapur, West Bengal, India. He is an Assistant Professor (Honorary) at the Academy of Scientific and Innovative Research, New Delhi, India. He is a Visiting Researcher at Hiroshima University Japan under Sakura Science Program funded by Japan Science and Technology Agency. He is a fellow of Consultancy Development Center, Govt. of India. He is a member of various scientific bodies including IEEE, Indian Science Congress Association, Institution of Engineers India and International Institute of Engineers. He is the member of the Governing council of "Nanoscience and Nanotechnology Society, India". He is a recipient of IEI-Young Engineers Award from Institution of Engineers India in Mechanical Engineering Discipline in 2019. His research areas include micro-nano systems engineering, controller development for micro machines and flexible electronics. He has pioneered development of indigenous micro fabrication machines named "Multi Fab" and "Nano Lase" in India and both the technologies are commercialized. He is also involved in skill development program under Govt of India, Skill Initiative and has conducted 15 training programs on Printed Circuit Board Manufacturing and Micromachining.

Prof. (Dr.) Nilotpal Ghosh, B.V.Sc.& A.H. (1st Class 1st & Gold Medalist), M.V.Sc. in APM (1st Class 1st), PhD, FNAPM, has 28 years of professional career in teaching, research and education administration. He is currently working as Professor of Livestock Production Management at West Bengal University of Animal & Fishery Sciences (WBUAFS), Kolkata. Prof. Ghosh started his service career as a Veterinary Officer, Animal Resources Development Department, Govt. of West Bengal, and also served as Professor & Head, Department of Animal Science, Bidhan Chandra Krishi Viswavidyalaya (BCKV), West Bengal, and Dean, Faculty of Veterinary & Animal Sciences, WBUAFS, Kolkata.He is associated with a number of scientific journals/magazines as a referee/scientific advisory member or a member of editorial board. He is a life member of a number of various professional societies in India. He has been the paper setter and external examiner of different universities in India and abroad. Fifty four research articles were in his credit, published in National and International journals. He has written enumerable popular articles in agriculture magazines/daily newspapers for awareness generation among the stake holders. He used to participate in the discourse in electronic media (television and radio) as an expert in livestock production. He has authored sixteen professional books which are acclaimed in the field of animal husbandry. His area of interest is livestock production management (especially dairy, poultry, goat and rabbit), human resource development of animal farming community and SHGs.

1

Dept. of Livestock Production Management, Faculty of Veterinary & Animal Sciences

View full text