Ingestive Behaviour Activities Based on Bioacoustic ...

DOI: 10.4018/IJAEIS.2020100105

International Journal of Agricultural and Environmental Information SystemsVolume 11 • Issue 4 • October-December 2020

Copyright©2020,IGIGlobal.CopyingordistributinginprintorelectronicformswithoutwrittenpermissionofIGIGlobalisprohibited.

69

Ingestive Behaviour Activities Based on Bioacoustic Signals in Grazing CattleGuilherme Augusto Defalque, College of Computing, Federal University of Mato Grosso do Sul, Brazil

https://orcid.org/0000-0002-7175-9992

Ricardo Santos, College of Computing, Federal University of Mato Grosso do Sul, Brazil

https://orcid.org/0000-0002-2955-8011

Victor Leonardo Yoshimura, College of Computing, Federal University of Mato Grosso do Sul, Brazil

https://orcid.org/0000-0002-4271-2140

Fabiana Villa Alves, Brazilian Agricultural Research Corporation / Embrapa Beef Cattle, Embrapa, Brazil

https://orcid.org/0000-0002-4244-940X

ABSTRACT

Thisworkfocusesonthedevelopmentofanautonomous,electronicplatformbasedonbioacoustictechniquestomonitoranimalsoundsrelatedtoingestiveactivitiesandsub-activitiesinbeefcattle.Theauthorshavedevelopedanelectronicplatformthatleveragetheusageofapplicationstopredicthealthandwelfareconditionsbasedonthebioacousticsignals.Theplatformarchitectureiscomprisedofelectronicsubsystemscoveringthesignalacquiringandfiltering;sampling,processing,datastoragesystem;andthecouplingoftheelectronicandcomputing.Theplatformiscoupledtotheanimal’shaltersothattheelectronicsystemisinacaseontheanimal’sforehead.Experimentswereperformedongrazingcattle(Nelorebreed)inthreefieldtests.Thesystemwasfullyabletoacquire,process,andstoretheanimal’sbioacousticsounds.Theauthorsusedanaudiofileprocessingsoftwaretoidentifyingestiveactivitiessuchasgrazing,rumination,andidleperiods.Theresults,basedonthesignalwaveform,achievedhighaccuracywhencomparedtothevisualmonitoringmethod.

KEywoRdSBeef Cattle, Bioacoustic Signals, Electronic Device, Grazing Cattle, Ingestive Behaviour, Nelore Breed, Non-Invasive, Welfare

INTRodUCTIoN

Marketdemandshave raised the levelof requirements concerningqualityof foodproducts, andspecifically,thequalityofanimalproducts(milkandmeat).Suchdemandsencourageresearchonfactorsthatinfluencebehaviourparametersthataredirectlyorindirectlyrelatedtoanimalhusbandry,suchasanimalwelfare.

The behaviour of an animal can be a clear indicator of its physiological state (Frost et al.,1997).Underthermalstress,cattlechangetheirbehaviourasawaytoregulatebodytemperatureandtomaintainhomothermia(Volpiet.al.,2020).Thesemechanismscanbephysiologicaland/orbehavioural,includingchangesiningestivebehaviourasawaytoovercomeadversesituations,whetherrelatedtoenvironment temperatureor to thequalityandquantityoffoodavailable.The

https://orcid.org/0000-0002-7175-9992

https://orcid.org/0000-0002-2955-8011

https://orcid.org/0000-0002-4271-2140

https://orcid.org/0000-0002-4244-940X


70

ingestiveactivities(grazingandrumination)canbeindicatorstheanimalhealth,welfareornutritionstatus(DeBoeveretal.,1990),oncethatananimalchangesitseatinghabitswhenithasitsroutinemodifiedduetophysicalchangesinthebodyortheenvironment.Monitoringingestivebehaviourisaprocedurethathelpsearlydetectionofdiseases,improvingwelfareandreducingeconomiclosses.

Onecommonlyadoptedproceduretomonitoringestivebehaviourincattleisthevisualmethod(Figure1,item2).Thismethodallowsexpertstoidentify,whilecattlegrazing,activitiesandjawmovementsrelatedtoingestivebehaviour.Thisprocessrequirestheobserverstobeatasignificantdistancefromtheanimalsinordernottodisrupttheirnaturalroutine.Expertsareresponsiblefortakingnotesoningestivebehaviourandanimal’senvironmentandpositioningdata(weather,temperature,animalposition,etc).Despitebeingawell-knownandsimpleprocedure,thevisualapproachlacksscaleandaccuracy,sinceitisatime-consuming,anderror-pronemethod,requiringanexperttobeonsite,monitoringtheanimal,andtakingnotesofitsactivities(Volpiet.al.,2020).

Thisworkpresentsanon-invasivedevicetomonitoringestivebehaviourbasedonthebioacoustictechnique(Figure1,item1).Thebioacousticstechniqueisatooltocharacterizeanimalbehaviourfromthesoundstheyproduce.Theno-vocalsoundsproducedbythejawwhenperformingactivitiesrelatedtoingestivebehaviourarethefocusofmanystudies(Volpiet.al.,2020),(Chelottiet.Al.,2016).Thosestudieshaveusedthebioacousticstechnique,visualmethod,andlabellingmethodforvalidationpurposes.

TheLabelingmethod(Figure1,item3)iscommonlyusedinbioacousticsdatatocharacterizeeventsandcompareresultsaswellasitworksasanimportanttoolinthedatasettrainingformachinelearning algorithms (Tullo et. al., 2017), (Chelotti, et. al., 2016). It consists in the observation,extraction, identification, and classification of individual animal sounds based on amplitude orfrequencyof thesoundsignal (Tullo,et.Al.,2017).Figure1presentsBPMN(BusinessProcessManagementNotation)diagramscorrelatingthebioacousticstechniquetothevisualmethodandlabelingmethod.Thebioacoustictechniqueusesthevisualmethodandlabellingmethodtoperformthevalidationprocessandtoserveasabasisforsounddatacapture.

Wehavedesignedanddevelopedalow-cost,non-invasivecomputationalandelectronicdevicefortheacquisition,processing,storage,andtransmissionofbioacousticssignalsingrazingcattle.WehaveevaluatedthedeviceingrazingNelore(Bos indicus)cattleandcomparedtheresultstothevisualtechniquetoidentifyingestivebehaviouractivities.Inadditiontotheflexibilityandautomaticdataacquisitionbroughtbythedevice,ourtestsshowedthattheproposedsolutionachievedhighaccuracycomparedtothevisualtechniqueondetectingactivitiessuchasruminating,idle,andgrazing.

RElATEd woRK

Inordertomeetscaleaccuracydemands,automatedsolutionsarebeingdevelopedtoreplacethevisualmethod.Someautomatedsolutionstodetectanimal’singestivebehaviourhasbeenbasedonaccelerometery,electromyography,pressuresensors,andbioacoustic(Andriamandrosoet.al.,2016).Amongthosetechniques,bioacoustichasshowntobeapromisingsupporttoolbyacquiringdataonfeedingparameters(bitingandchewingevents)toperformdetaileddataanalysisandtopredictcattlewelfarebasedongrazing,rumination,andidleactivities.

Mostoftheresearchworkonbioacousticsignalsfocusedonthedevelopmentofalgorithmstopredictbehaviourorproposinghardwarebasedoncommercialcomponents(recordsandmicrophones)(Volpiet.al.,2020),(Clapham,et.al.,2011),(Miloneet.Al.,2012),(Chelotti,et.al.,2016),(Denizet.al.,2017),(Laca&WallisDeVries,2000).Proposalsfocusingonoff-the-shelfcomponentsalongwiththevisualmethodwereusedtoidentifyingestiveactivitiesachievinganaccuracyofup95%(Claphamet.al.,2011),(Laca&WallisDeVries,2000).

Artificialintelligencetechniqueshavebeenusedinjawmovementsrecognitionsoftwaresforingestivebehaviour,suchasHiddenMarkovModels(Miloneet.al.,2012)andMachineLearning(Navonet.al.,2013)achievingaccuraciesof94%.Othersauthors((Navonet.al.,2013)and(Chelotti


71

et.al.,2016))adoptedtimedomaintechniquestoidentifyeventsthatcharacterizedjawmovementsoftheingestivebehaviouranddevelopedanalgorithmcapableofidentifyingandclassifyingjawmovementsofingestivebehaviourwithaccuracyof97.4%.Fewstudiesfocusedonthedevelopmentofdevices for ingestivebehaviourmonitoring.One themost recent studies (Denizet. al., 2017)aimedtodevelopaplatformwithalargesetoffunctionalitiesforreal-timemonitoringofingestivebehaviour.Thetestswereperformedinthelaboratory,whereadatabaseoflabeledsoundsofjawmovementsoftheingestivebehaviourwasreproducedprovidingaccuraciesonidentificationof92%andclassificationof78%.

Despitemanymethods,algorithms,andprototypes,thereisstillagapinproposalsfocusingondevicesandtechnologiesthatshowfeasibilityatfieldusage.Thefocusofthepreviousworkhavebeenonmonitoring ingestiveparametersondairycattle.Forgrazingbeefcattle, suchasNelorebreeds,viabletechnologicalsolutionsshouldbedurable,autonomous,andnon-invasivetomeettheanimal’sbehaviour.

ElECTRoNIC dEvICE

Wehavedesignedanelectronicdeviceforbioacousticssignalacquisitionthatisusedontheanimal’sforehead.Theelectronicdevicelocationisbasedonpreviouswork(Miloneet.al.,2012),(Denizet.al.,2017)and(Volpiet.al.,2020)indicatingthatingestivesoundscanpropagatealongtheskullthusbeingabletobeacquiredinapointoflessattenuation.TheelectronicdeviceiscomprisedofsubsystemswhichareprototypedinPrintedCircuitBoardsandstoredinacase,whichisontheanimal’sforehead,aspresentedinFigure2.

Figure 1. BPMN diagram of bioacoustic technique applied in grazing cattle


72

The subsystems are organized to Figure 3: acquiring and filtering system (subsystem 1);processing,storageandtransmissionsystem(subsystem2);andreceptionofbioacousticssignalssystem(subsystem3).

Acquisition and Filtering of Bioacoustics Signals Subsystem Thissubsystemacquiressignalsofmandibularsounds(Figure4(A),item5)usingamicrophonecapsule(A,1),wrappedinaninsulationmaterial(smoothacousticfoam).Thelowamplitudesignalsoftheelectretcapsuleareamplifiedbyacircuitinaninvertingtopology(A,2)andabiasDCcircuit

Figure 2. Electronic circuits are embedded into a case on the animal’s front head band

Figure 3. Block diagram of the bioacoustic device


73

(A,3)performsanoffsetsignalof2V,whichenablesthesignalstobereadbyananalog-to-digitalconverter.Theamplificationfactorisperformedbyadigitalpotentiometer(A,4)designedwithaCD4066IC(Figure4(B)).Theacousticfoampreventstheinfluenceofnoisescausedbywind,otherexternalnoises,andsoundscausedbytheanimalitself.Ananalogfilterisusedtominimizeunwantedsignals(noise).Thefilter(A,8)isafourth-orderButterworthtype,composedoftwosecond-orderfiltersincascadebySallen-Keytopology(items6-7inFigure4(A)).Thefiltercut-offfrequencyis2.6kHz.

Thefourresistors(R11,R5,R14,andR15)inFigure4(B),areenabledbytheselectionswitchesoftheCD4066IC(JP2andJP3inFigure4(B))bydigitalportscontainedinthemainboard(item7inFigure5),providingaminimumresistanceof1kOhm(approximateinternalresistanceofthe

Figure 4. Acquisition and filtering subsystem electronic schematic (A) and digital potentiometer design (B)


74

CD4066IC)upto200kOhms(whenalltheselectionswitchesaredisabled).Thepotentiometerisconnectedasshowninresistor6(Figure4(A),item4),attheinverterinputandattheoutputoftheoperationalamplifier,beingoneofthetoolstorealizethegainfactorofthemicrophone.

Processing and Transmission Subsystem Thissubsystem(Figure5)hasaninputforGPSVk2828u7g5lfmodule(5)capableofcollectinganimal’s location (latitude and longitude), a read and write module for micro SD cards (1) thatstoresGPSdataandbioacousticssoundsdata,aRealTimeClockDS1307IC(3),anATMega1284pmicrocontroller(4)fordataprocessing,amicrophoneinputforthemicrophonesystem(6)andanHC06Bluetoothtransmittermodule(2).AllthemodulesandICsarerepresentedbytheelectronicschematicinFigure5.

Thedevicepowersupplycircuit(Figure6)consistsofa1800mAh,3.7VLi-Ionbatteryoutput(4),abatterychargermoduleTP-4056(1)withmicroUSBinterfaceforbatterycharging,aboostmoduleMT3608(2)thatconvertsthebatteryvoltagefromaninputof3.7Vtoanoutputof7.2V,enoughtopowerupavoltageregulatormoduleLM7805(3),which(togetherwithtwocapacitors)maintainthesupplyvoltageoftheentiresystemat5V.

Bioacoustics Information and data Reception Subsystem Thissubsystemreceivesdatafromtheprocessingandtransmissionsubsystem(Figure5)encodedinPCM(PulseCodeModulation).ItisresponsibleforreceivingandstoringthesoundandGPSdatafilesinanindependentcomputersystem.ThesoftwarereceptionsystemwasdevelopedinPython3.6language.Figure7showsaflowchartofthecommunicationprotocolthathasbeendevelopedandtestedinlaboratory.

dEvICE PRoToTyPE

All subsystems were built on specific Printed Circuit Boards (PCBs) designed using AutodeskEagle™software.ThePCBsarestoredintoacasebasedonPolyLacticAcid(PLA)material.Figure8(A)presentsallthetechnologicalapparatusdesignedinsidethecaseandprotectedwithsmoothacousticfoam.

Thecaseisplacedontheanimal’sforeheadusingacustom-madehalter(Figure8(B))tofixthecaseontotheanimal’sforeheadandneck.Thehalterhasatapethatallowsthecasetobefixedtothecattleforehead(Figure8(B),item1),atapethatsurroundsthelowerpartoftheanimal’sneck

Figure 5. Electronic components for sampling, processing, and communication


75

(2)andatapeforattachmenttothecontouroftheneck(4).Inaddition,aleatherband(3)passesovertheanimal’shead.Thecasehas3.82incheslength,3.43incheswidth,3.54inchesheightandaweightof350grams.

Table 1 shows the maximum continuous consumption (mA) for all subsystems. In DatatransmissionandLocalizationSubsystems(GPSandBluetooth)thedirectcurrentofthelocationacquisitionsystem(GPS)is32mAwhendataisbeingacquiredandthemaximumconsumptionofthetransmissionsystemis30-40mA,however,withoutdatatransmission,itsconsumptionisaround8mA;inthestorageandprocessingsubsystem,themicroSDCardModuleconsumes200mAwheninwritemode.AcquisitionandFilteringSubsystemconsumesmaximumcontinuousof3.6mA,

Figure 6. Power supply system electronic schematic

Figure 7. Flowchart of communication between the receiver software and the platform


76

where,TwoLM324areresponsibleforconsumingmostofthecontinuouscurrent(2.4mA).PowerSupplySubsystemconsumes8.3mAandLM7805isthecomponentthatconsumespracticallyallthecurrentofthesubsystem(8mA).

ExPERImENTS ANd RESUlTS

ExperimentswerecarriedoutattheHighPerformanceComputingSystemsLaboratory(LSCAD)oftheFederalUniversityofMatoGrossodoSul(UFMS)andatEMBRAPABeefCattletoobserveandanalysetheplatformperformanceandsuitabilitytobeusedbyNeloreanimals.Thelaboratoryexperimentstook20hoursandgeneratedabout300MBofsounddata.Experimentsandtestatfieldtookapproximately9hoursgeneratedabout254MBofnewsounddata.

Inordertotestandvalidatethedevice,weperformedexperimentsinthelaboratoryusingalabelleddatasetfromgrazing,rumination,andidleactivities(Volpiet.al.,2020).Thedatasetconsistsofbioacousticsingestivedatafrom12NelorecowsgrazingatapastureofBrachiaria brizantha.Eachanimalwas30monthsoldandhadanaverageliveweightof400kg.Eachanimalusedagenericlapel

Figure 8. Electronic device case and and halter

Table 1. Main components and maximum current consumption

Component Components Maximum continuous total consumption (mA)

AcquisitionandFilteringSubsystem

LM358TwoLM324

CD40663.6

PowerSupplySubsystemTP-4056MT3608LM7805

8.3

DataTransmissionandLocalizationAcquisition

Subsystems

HC06GPS–

VK2828U7G5LF75

Processinganddatastoresubsystem

MicroSDCardModuleATMega1284pRTC–DS1307

LED5mm

225.5


77

microphoneanddigitalvoicerecorder(Sony,100ICD-PX240™).Afterrecordingthesounddata,expertslabeledthecollecteddatacomparingthebioacousticssignalstothevisualmethodannotations.

Weperformedanexperimenttoevaluateifourdevicecouldacquiresounddatafromthedatasetandtocomparethesignalswaveforms.Theexperimentwasreproducedinanacousticallyisolatedroom,theaudiofilesrepresentingtheactivitiesoftheingestivebehaviourwerevisuallycomparedtotheexcerptsfromthedatasetusingtheOcenaudio™software.Eventhoughthedatafromthedatasetwereacquiredwithdifferentparametersforacquisitionandsampling(44kHzsamplingwasusedinthedatasetrecorder)comparedtoourdevice,theresultsshowedvisualandauditorysimilarity.

Another set of experiments were performed at the field, in a grazing paddock with water,forageanddrymineralsavailable.TheseexperimentsusedtheplatformshowninFigure8.ThedataacquisitionworkflowwasperformedaccordingtoFigure7.Thedataacquisitiontimeintervalwassetupbeforethestartofeachexperiment.ThetransmissionsystemwasnotenabledatfieldtestssothatallsounddatawerestoredandavailableintothemicroSDcard.

Figure9showsimagesofthefieldtestscarriedoutattheintheAgriculturalResearchCorporation(EMBRAPABeefCattle).Animalswere20monthsold(Neloreheifers).Table2presentsthedate,time,batteryconsumptionlevel,sounddatarecordsetupandthemainresultsofeachtest.

AspresentedinTable2,thefirsttestwasperformedonOctober26th,2018.Itallowedforfine-tuninginthehalter,circuits,andcase.Thesecondtest(November26th,2018)showednormalcircuitoperationinsoundcaptureanddatastorage.ThetestonDecember21st,2018,targetedplatformvalidation so that an expert, using the visual method, was taking notes on ingestive behaviour

Figure 9. Animal using the bioacoustics device while grazing

Table 2. Summary of the field tests

Date Hour Battery (V) Sound data setup Results

October26th,2018 08:55-09:17a.m. 4.12–4.06

1minuteperfile,3minutesinterval

betweenrecordings

Halterandcircuitcasevalidation

November30th,2018 08:38-10:35a.m. 4.11–4.02

Normalcircuitoperationforsoundacquisitionand

datastorage

December21st,2018 08:15-02:07p.m. 4.20–3.89

2minutesperfile,3minutesinterval,betweenrecordings

Acquisitionandstoragedataandcomparisonofbioacousticsdatawithanalysismadethrough

visualobservationmethod


78

(rumination,grazing,andidle)fromoneanimalevery10minutes.Someextrasamplestothesamplingperiodwerecollected,asweaimedtocollectasmanysamplesaspossiblefromthethreeactivitiesofingestivebehaviour.WeperformedavisualandauditoryanalysisoftherecordedsoundsandcomparedtheactivitiesoftheingestivebehaviourtothevisualmethodusingtheOcenaudio™software.

Figures10and11show,respectively,soundpatternsofgrazingandruminationactivities.Figure12presentssignalfoundwhentheanimalperformssomeothertypeofactivity(idle).InFigures10-12theverticalaxisrepresentstheanaloguesignalwaveformandtheHertzvaluesoffrequencies.Thehorizontalaxisrepresentstheelapsedtimeinmiliseconds.

Theanalysesof soundsamples inFigure10showed thatgrazingevents ranged from0.4 to1seconds(incontinuousstreamofactivity),whereeacheventhadhighersignal intensities thanrumination.Variationsonsignalintensityoccurduetothebiteevents(whichareresponsibleforsignalswithgreateramplitudes),chewevents(signalswithoutsteep,moderatepeaks)andchew-biteevents,acombinationofbothmovements.Grazingsignalsshowedhigherenergypeaksatfrequenciesupto4kHzwhichcorroboratetosomepreviousscientificstudies(Volpiet.al.,2020),(adriamandrosoet.al.,2016).

Ruminationactivitywascharacterizedwithalowamplitudeofsignals.Thesignalswithlowamplitudecanbeexplainedconsideringthatachewingofregurgitatedforageoccurs,requiringalowerchewingintensitycomparedtothegrazingactivity.Ruminationeventswereidentifiedatintervalsfrom0.5upto0.8seconds.Eacheventhad0.15to0.35seconds.Thefrequencieswithhigherintensitieswerecharacterizedcloseto2kHz.

Table 3 presents a summary of comparing the manual labeling with the bioacoustic signalidentification from our device. The results are from the samples acquired in the experiment onDecember21st,2018.

The periods where the device results indicate “Device was idle’” were attempts to acquirelocationdatausingGPSorbyprogrammedidleperiodsforpowersaving.Consideringtheperiodsthattheplatformperformedsignalsacquisition(disregardingtheperiodswheretheplatformwasidle),comparedtothevisualmethod,thedataacquiredfromthedevicepresentedhighaccuracyforeach

Figure 10. Acoustic signals from the grazing activity


79

activityoftheingestivebehaviour(grazing,rumination,andidle).Wecalculatethestandarddeviation,samplemean,andvarianceofthesampledatafromtherumination,idle,andgrazingdatasets.Thevarianceofthegrazingdatasetwas2.47,ruminationwas0.43,andidle0.27.Thestandarddeviationforalldatasetswere1.57(grazing),0.66(rumination),and0.27(idle).Thesamplemeanwas102.8(grazing),and102.9(idle).Sincetherearealargernumberofsamplesineachdataset(largerthan1000000samples),themeandifferencewasnotsignificantamongtheacitivies.

Figure 11. Acoustic signals from the rumination activity

Figure 12. Acoustic signals from the idle activity


80

CoNClUSIoN ANd FUTURE woRK

Thisworkpresentsanon-invasivedevicetoacquirebioacousticssignalsfromgrazingcattle.Thedevicecanbeusedasasupporttoolonexperimentstoidentifyingestivebehaviouractivitiessuchasidle,rumination,andgrazingfrombioacousticssignals.

Theexperimentsinlaboratoryaimedatidentifyingbioacousticswaveformpatternsfromeachingestiveactivityandimprovingthedesignofthedatatransmissionsystem.Theexperimentsongrazingcattleatfiledindicatedthatthedeviceseemssuitableforcattleusage,evenforlong-termexperiments.Wehaveusedthedevicelongerthan6uninterruptedhoursatfieldunderenvironmentalconditionssuchassunnyandrainyweather.Laboratorytestswerealsoperformedwasincontinuousoperationpresentinganautonomyofapproximately15hours.Theplatformreplacementtolowenergyconsumptionmoduleisoneoftheactivitiesweintendtotakeinaccounttoimprovetheplatformenergyefficienty.

Tothebestofourknowledge,theproposedelectronicplatformisafirstsolutiononautomaticbioacousticssignalsacquisitionabletobeusedasalong-termtoolappliedtograzingcattle.Theelectronicdevicecanbeusedbyapplicationsfocusingonautomaticallyidentifyingestiveactivities

Table 3. Comparison between the visual method and the bioacoustic device results

Hour Note Visual Method Bioacoustic Method

11:00a.m. AWSS Idle Idle



11:12a.m. AWSS Idle Devicewasidle


11:26a.m. AWSBC Grazing Grazing


11:40a.m. AWSS Grazing Grazing

11:50a.m. AWSS Grazing Grazing

12:00p.m. AWSS Idle Idle



12:20p.m. ALDSW Rumination Rumination




12:50p.m. ALDSW Idle Devicewasidle

01:00p.m. ALDSW Idle Devicewasidle

01:10p.m. ALDSW Idle Idle


01:24p.m. AWSS Grazing Grazing



AWSS = Animal was standing, sunny weather; AWSBC = Animal was standing, sunny weather and clouds; ALDSW = Animal lying down, sunny weather.


81

andsubactivitiesfrombioacousticssignals.Itcanbealsousedasasupporttooltoaiddecision-makersonprovidingmoreaccurateinformationoningestivebehaviour,allowingforpreciseanimalmonitoring,thatcanoptimizedietsandothermanagementpracticesthatleadtoimprovedanimalwelfareandperformance.

ACKNowlEdGmENT

TheauthorsthankBrazilianResearchAgenciesFUNDECT,CAPES,andCNPq,fortheirfinancialsupporttothiswork.TheyalsothanktheirinstitutionsFederalUniversityofMatoGrossodoSul(UFMS)andEMBRAPABeefCattlefortheirinfrastructureandlogisticssupporttothiswork.


82

REFERENCES

Andriamandroso,A.,Bindelle,J.,Mercatoris,B.,&Lebeau,F.(2016).AReviewontheUseofSensorstoMonitorCattleJawMovementsandBehaviorwhenGrazing.Biotechnologie, Agronomie, Société et Environnement,20.

Chelotti,J.O.,Vanrell,S.R.,Milone,D.H.,Utsumi,S.A.,Galli,J.R.,Rufiner,H.L.,&Giovanini,L.L.(2016).AReal-timeAlgorithmforAcousticMonitoringofIngestiveBehaviorofGrazingCattle.Computers and Electronics in Agriculture,127,64–75.doi:10.1016/j.compag.2016.05.015

Clapham,W.M.,Fedders,J.M.,Beeman,K.,&Neel,J.P.(2011).AcousticMonitoringSystemtoQuantifyIngestiveBehaviorofFree-grazingCattle.Computers and Electronics in Agriculture,76(1),96–104.doi:10.1016/j.compag.2011.01.009

DeBoever,J.L.,Andries,J.I.,DeBrabander,D.L.,Cottyn,B.G.,&Buysse,F.X.(1990).ChewingActivityofRuminantsasaMeasureofPhysicalStructure—AReviewofFactorsAffectingit.Animal Feed Science and Technology,27(4),281–291.doi:10.1016/0377-8401(90)90143-V

Deniz,N.N.,Chelotti,J.O.,Galli,J.R.,Planisich,A.M.,Larripa,M.J.,Rufiner,H.L.,&Giovanini,L.L.(2017).EmbeddedSystemforReal-timeMonitoringofForagingBehaviorofGrazingCattleusingAcousticSignals.Computers and Electronics in Agriculture,138,167–174.doi:10.1016/j.compag.2017.04.024

Frost,A.R.,Schofield,C.P.,Beaulah,S.A.,Mottram,T.T.,Lines,J.A.,&Wathes,C.M.(1997).AReviewofLivestockMonitoringandtheNeedforIntegratedSystems.Computers and Electronics in Agriculture,17(2),139–159.doi:10.1016/S0168-1699(96)01301-4

Laca,E.A.,&WallisDeVries,.(2000).AcousticMeasurementofIntakeandGrazingBehaviourofCattle.Grass and Forage Science,55(2),97–104.doi:10.1046/j.1365-2494.2000.00203.x

Milone,D.H.,Galli,J.R.,Cangiano,C.A.,Rufiner,H.L.,&Laca,E.A.(2012).AutomaticRecognitionofIngestiveSoundsofCattlebasedonHiddenMarkovmodels.Computers and Electronics in Agriculture,87,51–55.doi:10.1016/j.compag.2012.05.004

Navon,S.,Mizrach,A.,Hetzroni,A.,&Ungar,E.D.(2013).AutomaticRecognitionofJawMovementsinFree-rangingCattle,GoatsandSheep,usingAcousticMonitoring.Biosystems Engineering,114(4),474–483.doi:10.1016/j.biosystemseng.2012.08.005

Tullo,E.,Fontana,I.,Diana,A.,Norton,T.,Berckmans,D.,&Guarino,M.(2017).ApplicationNote:Labelling,aMethodologytoDevelopReliableAlgorithminPLF.Computers and Electronics in Agriculture,142,424–428.doi:10.1016/j.compag.2017.09.030

Volpi,D.,Vigas,V.P.,Alves,F.V.,Silva,G.O.,&Saraiva,E.F.(2020).AnApplicationofHotelling’sT2TestfortheComparisonoftheVisual-acousticMethodintheIdentificationofIngestiveCattleBehaviour.Brazilian Journal of Biometrics,38(1),79–91.

http://dx.doi.org/10.1016/j.compag.2016.05.015



http://dx.doi.org/10.1016/0377-8401(90)90143-V


http://dx.doi.org/10.1016/S0168-1699(96)01301-4

http://dx.doi.org/10.1046/j.1365-2494.2000.00203.x


http://dx.doi.org/10.1016/j.biosystemseng.2012.08.005



83

Guilherme Defalque was born in Buritama, São Paulo, Brazil on May 7th, 1991. He obtained his Computer Engineering (UFMS), Master in Computer Engineering (UFMS) and is currently a doctoral student in computer science (UFMS). His main research interests are digital circuit designs, precision livestock, IOT and sensors design.Ricardo Ribeiro Santos is with the College of Computing of the Federal University of Mato Grosso do Sul. He got a PhD in computer science from the Institute of Computing of the State University of Campinas (UNICAMP), in 2007. He was a visiting scholar at the Department of Civil and Environmental Enginnering of the Stanford University in 2018.

Victor Leonardo Yoshimura was born in Campo Grande, Mato Grosso do Sul, Brazil on May 3rd, 1977. He obtained his Electrical Engineer, Master in Electrical Engineering and Doctor in Electrical Engineering degrees from UFMS (1999), UFSC (2002) and UNESP (2013), respectively. He also holds a degree in Mathematics from UFMT (2009). Currently, he is with Faculdade de Computação (UFMS) and his main research interests are LMI-based control designs and applications for electronic circuits.

Fabiana Villa Alves has a degree in Animal Science from Universidade Federal de Lavras (2002) and Ph.D. in Animal Science and Pastures from Escola Superior de Agricultura Luiz de Queiroz - ESALQ/USP (2007). Research focus is Sustainable Cattle Production with particular emphasis on animal welfare and thermal comfort under integrated systems. She is senior researcher at the Brazilian Agricultural Research Corporation - Embrapa Beef Cattle, in Campo Grande (MS), where she is also deputy supervisor for the Research Group on Production Systems (GSP). She collaborates on several national and international projects on her subject, being creator and leader of the recent and innovative public-private initiative “Carbon Neutral Brazilian Beef”. Having an extensive network with scientists, policy makers and the industry all over Brazil and many parts of the world, she is member of several committees, being the official representative for Brazil on FAO’s Global Agenda for Sustainable Livestock (GASL) and Livestock Environmental Assessment and Performance Partnership (LEAP).

Ingestive Behaviour Activities Based on Bioacoustic ...

Documents

Transcript of Ingestive Behaviour Activities Based on Bioacoustic ...