TinyMachineLearning(TinyML)isafieldofmachinelearningthatfocusesonthedevelopmentanddeploymentofMLmodelsonlowpower,low-footprintmicrocontrollerdeviceslikeanArduino,forexample.
Machinelearningisafieldofcomputersystemsthataimstodevelopself-improvingalgorithmsandstatisticalmodels.Thisisdonewithmassiveamountsofdata,whichamodelanalysesandextractspatternsfrominordertolearnandimproveonagiventask.Thisseeminglysimpleparadigmhasledtogroundbreakingadvancementsincomplextaskslikeforecasting,anomalydetection,andcomputervision!
Ourdevicesisnowsupportedbyaverylargenumberofco-operativeplatformsandcases.
Bysupportingtheseplatforms,weenabledeveloperstoeasilybuildanddeploymachinelearningmodelsonedgedevices,frommicrocontrollerstosingle-boardcomputers.
SenseCraftModelAssistantisanopen-sourceprojectfocusedonembeddedAI,developedbySeeedStudio.Itoffersarangeofoptimizedalgorithmsforreal-worldscenarios,makingimplementationmoreuser-friendlyandachievingfasterandmoreaccurateinferenceonembeddeddevices.
SenseCraftModelAssistantcurrentlysupportsanomalydetection,computervision,andscenario-specificalgorithms,withmoretobeaddedinthefuture.
Itprovidesauser-friendlyplatformfortrainingoncollecteddataandvisualizingalgorithmperformance,anditsmodelsaredesignedtorunonlow-costhardwaresuchasESP32,Arduinodevelopmentboards,andRaspberryPi.
SenseCraftModelAssistantalsosupportsmultipleformatsformodelexport,includingTensorFlowLite,ONNX,andspecialformatslikeTensorRTandOpenVINO.WithSenseCraftModelAssistant,developerscaneasilybuildanddeploymachinelearningmodelsonawiderangeofembeddeddevices.
WithCodecraftandWioTerminal,itisnowpossibletoexperiencetheentireprocessofembeddedmachinelearningwithouthavingtodealwithacomplexprogrammingenvironmentandprogrammingknowledge.
PoweredbyEdgeImpulse,TinyMachineLearningiseasilyaccessiblebybeginnersusingCodecraftgraphicalprogramming.Bysimpledrag-and-dropcoding,acquiringdata,training,anddeployingmodelismorevividthanever.
EdgeImpulseisapowerfulmachinelearningplatformforbuildinganddeployingembeddedmachinelearningmodels.
Itprovidesdeveloperswitharangeoftoolsforcollectingandprocessingsensordata,designingandtrainingmachinelearningmodels,anddeployingthosemodelstoedgedevices.
EdgeImpulseisoptimizedforsensor-basedapplicationsandsupportsawiderangeofhardwareplatforms.WithEdgeImpulse,developerscaneasilybuildanddeploymachinelearningmodelsonembeddeddevices,makingitavaluabletoolforIoTandsmartdeviceapplications.
TensorFlowLiteisalightweightversionofthepopularTensorFlowmachinelearningframework,designedforrunningonembeddedandmobiledevices.
Itprovidesdeveloperswitharangeoftoolsforbuildinganddeployingmachinelearningmodelsonresource-constraineddevices,includingsupportforspecializedhardwareaccelerators.
TensorFlowLitesupportsawiderangeofhardwareplatformsandisparticularlywell-suitedfordeeplearningapplications.WithTensorFlowLite,developerscaneasilybuildanddeploymachinelearningmodelsonembeddedandmobiledevices,makingitavaluabletoolforawiderangeofapplicationsintheIoT,mobile,andsmartdevicedomains.
WioTerminalisapowerfulandeasy-to-usedevelopmentboarddesignedformakers,hobbyists,andIoTenthusiasts.Itfeaturesa2.4-inchLCDscreen,Wi-FiandBluetoothconnectivity,anarrayofsensors,andavarietyofinput/outputinterfaces.
WioTerminalsupportsarangeofprogramminglanguagesincludingArduino,MicroPython,CircuitPython,andmore,makingiteasyfordeveloperstogetstarted.Withitscompactandportabledesign,WioTerminalisidealforawiderangeofprojects,fromcreatingsmartdevicestobuildingprototypesforindustrialapplications.
HerewewillshowyousomegreatexamplesofourTinyMLdevices.Thesecasestudieswillbecategorisedbyapplicationscenarios,sowehopeyoucanfindsomethingofinteresthereandjoinusinrealisingyourideas!
UsingTinyMLtechnologyitispossibletocombinesensordatawithmachinelearningmodelstoenablehuman-computerinteractionapplicationssuchasspeechrecognition,gesturerecognitionandposerecognition.
TinyMLtechnologycanbeappliedintheretailindustryforsalesforecasting,customerbehaviouranalysis,resourceoptimisation,losspreventionandsmartmarketing,helpingtoachievesmarterandmoreefficientbusinessmanagementandoperations.
UsingTinyMLtechnologyitispossibletocombinesensordatawithmachinelearningmodelstoautomateirrigation,weatherforecasting,andcropdiseaseandpestdetection.
UsingTinyMLtechnologyitispossibletocombinesensordatawithmachinelearningmodelstoautomateandoptimiseproductionprocessessuchasqualitycontrolandequipmentmaintenance.
TinyMLcanbeappliedtohealthmonitoringtohelpthemedicalindustryachievemoreefficientandaccuratehealthmonitoring.
Thankyouforchoosingourproducts!Weareheretoprovideyouwithdifferentsupporttoensurethatyourexperiencewithourproductsisassmoothaspossible.Weofferseveralcommunicationchannelstocatertodifferentpreferencesandneeds.