Facial recognition system - Wikipedia

A facial recognition system is a technology capable of matching a human face from a digital image or a video frame against a database of faces, ... Facialrecognitionsystem FromWikipedia,thefreeencyclopedia Jumptonavigation Jumptosearch Thisarticle'sfactualaccuracymaybecompromisedduetoout-of-dateinformation.Pleasehelpupdatethisarticletoreflectrecenteventsornewlyavailableinformation.(June2022) Technologycapableofmatchingafacefromanimageagainstadatabaseoffaces "Facerecognition"redirectshere.Forthehumancognitiveprocess,seefaceperception.Forotheruses,seefacialrecognition. AutomaticticketgatewithfacerecognitionsysteminOsakaMetroMorinomiyaStation Afacialrecognitionsystemisatechnologycapableofmatchingahumanfacefromadigitalimageoravideoframeagainstadatabaseoffaces,typicallyemployedtoauthenticateusersthroughIDverificationservices,worksbypinpointingandmeasuringfacialfeaturesfromagivenimage.[1] Developmentbeganonsimilarsystemsinthe1960s,beginningasaformofcomputerapplication.Sincetheirinception,facialrecognitionsystemshaveseenwiderusesinrecenttimesonsmartphonesandinotherformsoftechnology,suchasrobotics.Becausecomputerizedfacialrecognitioninvolvesthemeasurementofahuman'sphysiologicalcharacteristics,facialrecognitionsystemsarecategorizedasbiometrics.Althoughtheaccuracyoffacialrecognitionsystemsasabiometrictechnologyislowerthanirisrecognitionandfingerprintrecognition,itiswidelyadoptedduetoitscontactlessprocess.[2]Facialrecognitionsystemshavebeendeployedinadvancedhuman–computerinteraction,videosurveillanceandautomaticindexingofimages.[3] Facialrecognitionsystemsareemployedthroughouttheworldtodaybygovernmentsandprivatecompanies.[4]Theireffectivenessvaries,andsomesystemshavepreviouslybeenscrappedbecauseoftheirineffectiveness.Theuseoffacialrecognitionsystemshasalsoraisedcontroversy,withclaimsthatthesystemsviolatecitizens'privacy,commonlymakeincorrectidentifications,encouragegendernormsandracialprofiling,anddonotprotectimportantbiometricdata.Theappearanceofsyntheticmediasuchasdeepfakeshasalsoraisedconcernsaboutitssecurity.[5]TheseclaimshaveledtothebanoffacialrecognitionsystemsinseveralcitiesintheUnitedStates.[6]Asaresultofgrowingsocietalconcerns,Metaannounced[7]thatitplanstoshutdownFacebookfacialrecognitionsystem,deletingthefacescandataofmorethanonebillionusers.[8]Thischangewillrepresentoneofthelargestshiftsinfacialrecognitionusageinthetechnology'shistory. Contents 1Historyoffacialrecognitiontechnology 2Techniquesforfacerecognition 2.1Traditional 2.2Humanidentificationatadistance(HID) 2.33-dimensionalrecognition 2.4Thermalcameras 3Application 3.1Socialmedia 3.2IDverification 3.2.1FaceID 3.3DeploymentofFRTforavailinggovernmentservices 3.3.1India 3.4Deploymentinsecurityservices 3.4.1Commonwealth 3.4.2UnitedStates 3.4.3China 3.4.4India 3.4.5LatinAmerica 3.4.6EuropeanUnion 3.4.6.1Greece 3.4.6.2Italy 3.4.6.3TheNetherlands 3.4.7SouthAfrica 3.5Deploymentinretailstores 3.5.1UnitedStates 3.5.2Australia 3.6Additionaluses 4Advantagesanddisadvantages 4.1Comparedtootherbiometricsystems 4.2Weaknesses 4.3Ineffectiveness 5Controversies 5.1Privacyviolations 5.2Imperfecttechnologyinlawenforcement 5.3Dataprotection 6Bansontheuseoffacialrecognitiontechnology 7Emotionrecognition 8Anti-facialrecognitionsystems 9Seealso 10References 11Furtherreading 12Externallinks Historyoffacialrecognitiontechnology[edit] Automatedfacialrecognitionwaspioneeredinthe1960s.WoodyBledsoe,HelenChanWolf,andCharlesBissonworkedonusingthecomputertorecognizehumanfaces.Theirearlyfacialrecognitionprojectwasdubbed"man-machine"becausethecoordinatesofthefacialfeaturesinaphotographhadtobeestablishedbyahumanbeforetheycouldbeusedbythecomputerforrecognition.Onagraphicstabletahumanhadtopinpointthecoordinatesoffacialfeaturessuchasthepupilcenters,theinsideandoutsidecornerofeyes,andthewidowspeakinthehairline.Thecoordinateswereusedtocalculate20distances,includingthewidthofthemouthandoftheeyes.Ahumancouldprocessabout40picturesanhourinthismannerandsobuildadatabaseofthecomputeddistances.Acomputerwouldthenautomaticallycomparethedistancesforeachphotograph,calculatethedifferencebetweenthedistancesandreturntheclosedrecordsasapossiblematch.[9] In1970,TakeoKanadepubliclydemonstratedaface-matchingsystemthatlocatedanatomicalfeaturessuchasthechinandcalculatedthedistanceratiobetweenfacialfeatureswithouthumanintervention.Latertestsrevealedthatthesystemcouldnotalwaysreliablyidentifyfacialfeatures.Nonetheless,interestinthesubjectgrewandin1977Kanadepublishedthefirstdetailedbookonfacialrecognitiontechnology.[10] In1993,theDefenseAdvancedResearchProjectAgency(DARPA)andtheArmyResearchLaboratory(ARL)establishedthefacerecognitiontechnologyprogramFERETtodevelop"automaticfacerecognitioncapabilities"thatcouldbeemployedinaproductivereallifeenvironment"toassistsecurity,intelligence,andlawenforcementpersonnelintheperformanceoftheirduties."FacerecognitionsystemsthathadbeentrialedinresearchlabswereevaluatedandtheFERETtestsfoundthatwhiletheperformanceofexistingautomatedfacialrecognitionsystemsvaried,ahandfulofexistingmethodscouldviablybeusedtorecognizefacesinstillimagestakeninacontrolledenvironment.[11]TheFERETtestsspawnedthreeUScompaniesthatsoldautomatedfacialrecognitionsystems.VisionCorporationandMirosIncwerebothfoundedin1994,byresearcherswhousedtheresultsoftheFERETtestsasasellingpoint.ViisageTechnologywasestablishedbyaidentificationcarddefensecontractorin1996tocommerciallyexploittherightstothefacialrecognitionalgorithmdevelopedbyAlexPentlandatMIT.[12] Followingthe1993FERETface-recognitionvendortesttheDepartmentofMotorVehicles(DMV)officesinWestVirginiaandNewMexicowerethefirstDMVofficestouseautomatedfacialrecognitionsystemsasawaytopreventanddetectpeopleobtainingmultipledrivinglicensesunderdifferentnames.Driver'slicensesintheUnitedStateswereatthatpointacommonlyacceptedformofphotoidentification.DMVofficesacrosstheUnitedStateswereundergoingatechnologicalupgradeandwereintheprocessofestablishingdatabasesofdigitalIDphotographs.ThisenabledDMVofficestodeploythefacialrecognitionsystemsonthemarkettosearchphotographsfornewdrivinglicensesagainsttheexistingDMVdatabase.[13]DMVofficesbecameoneofthefirstmajormarketsforautomatedfacialrecognitiontechnologyandintroducedUScitizenstofacialrecognitionasastandardmethodofidentification.[14]TheincreaseoftheUSprisonpopulationinthe1990spromptedU.S.statestoestablishedconnectedandautomatedidentificationsystemsthatincorporateddigitalbiometricdatabases,insomeinstancesthisincludedfacialrecognition.In1999,MinnesotaincorporatedthefacialrecognitionsystemFaceITbyVisionicsintoamugshotbookingsystemthatallowedpolice,judgesandcourtofficerstotrackcriminalsacrossthestate.[15] Inthisshearmappingtheredarrowchangesdirection,butthebluearrowdoesnotandisusedaseigenvector. TheViola–JonesalgorithmforfacedetectionusesHaar-likefeaturestolocatefacesinanimage.HereaHaarfeaturethatlookssimilartothebridgeofthenoseisappliedontotheface. Untilthe1990s,facialrecognitionsystemsweredevelopedprimarilybyusingphotographicportraitsofhumanfaces.Researchonfacerecognitiontoreliablylocateafaceinanimagethatcontainsotherobjectsgainedtractionintheearly1990swiththeprinciplecomponentanalysis(PCA).ThePCAmethodoffacedetectionisalsoknownasEigenfaceandwasdevelopedbyMatthewTurkandAlexPentland.[16]TurkandPentlandcombinedtheconceptualapproachoftheKarhunen–Loèvetheoremandfactoranalysis,todevelopalinearmodel.Eigenfacesaredeterminedbasedonglobalandorthogonalfeaturesinhumanfaces.AhumanfaceiscalculatedasaweightedcombinationofanumberofEigenfaces.BecausefewEigenfaceswereusedtoencodehumanfacesofagivenpopulation,TurkandPentland'sPCAfacedetectionmethodgreatlyreducedtheamountofdatathathadtobeprocessedtodetectaface.Pentlandin1994definedEigenfacefeatures,includingeigeneyes,eigenmouthsandeigennoses,toadvancetheuseofPCAinfacialrecognition.In1997,thePCAEigenfacemethodoffacerecognition[17]wasimproveduponusinglineardiscriminantanalysis(LDA)toproduceFisherfaces.[18]LDAFisherfacesbecamedominantlyusedinPCAfeaturebasedfacerecognition.WhileEigenfaceswerealsousedforfacereconstruction.Intheseapproachesnoglobalstructureofthefaceiscalculatedwhichlinksthefacialfeaturesorparts.[19] Purelyfeaturebasedapproachestofacialrecognitionwereovertakeninthelate1990sbytheBochumsystem,whichusedGaborfiltertorecordthefacefeaturesandcomputedagridofthefacestructuretolinkthefeatures.[20]ChristophvonderMalsburgandhisresearchteamattheUniversityofBochumdevelopedElasticBunchGraphMatchinginthemid-1990stoextractafaceoutofanimageusingskinsegmentation.[21]By1997,thefacedetectionmethoddevelopedbyMalsburgoutperformedmostotherfacialdetectionsystemsonthemarket.Theso-called"Bochumsystem"offacedetectionwassoldcommerciallyonthemarketasZN-Facetooperatorsofairportsandotherbusylocations.Thesoftwarewas"robustenoughtomakeidentificationsfromless-than-perfectfaceviews.Itcanalsooftenseethroughsuchimpedimentstoidentificationasmustaches,beards,changedhairstylesandglasses—evensunglasses".[22] Real-timefacedetectioninvideofootagebecamepossiblein2001withtheViola–Jonesobjectdetectionframeworkforfaces.[23]PaulViolaandMichaelJonescombinedtheirfacedetectionmethodwiththeHaar-likefeatureapproachtoobjectrecognitionindigitalimagestolaunchAdaBoost,thefirstreal-timefrontal-viewfacedetector.[24]By2015,theViola–Jonesalgorithmhadbeenimplementedusingsmalllowpowerdetectorsonhandhelddevicesandembeddedsystems.Therefore,theViola–Jonesalgorithmhasnotonlybroadenedthepracticalapplicationoffacerecognitionsystemsbuthasalsobeenusedtosupportnewfeaturesinuserinterfacesandteleconferencing.[25] UkraineisusingfacialrecognitiontoidentifydeadRussiasoldiers.Ukrainehasconducted8,600searchesandidentifiedthefamiliesof582deceasedRussiansoldiers.HavebeendonebytheITvolunteersectionoftheUkrainianarmy.UkraineisusingtheUS-basedClearviewAIsoftware.ThemainaimistodestabilisetheRussiangovernment.Asaformofpsychologicalwarfare.Some340Ukrainiangovernmentofficialsinfivegovernmentministriesareusingthetechnology.ItisusedtocatchspiesthatmighttrytoenterUkraine.[26] ClearviewAI’sfacialrecognitiondatabaseisonlyavailabletogovernmentagencieswhomayonlyusethetechnologytoassistinthecourseoflawenforcementinvestigationsorinconnectionwithnationalsecurity.[27] ThesoftwarewasdonatedtoUkrainebyClearviewAI.Russiaisthoughttobeusingittofindanti-waractivists.OriginallydesignedforUSlawenforcement.Usingitinwardeadraisesnewconcerns.OneLondonbasedsurveillanceexpert,StephenHare,hasitmightmaketheUkrainiansappearinhuman:"Isitactuallyworking?Orisitmaking[Russians]say:'Lookattheselawless,cruelUkrainians,doingthistoourboys?'"[28] Techniquesforfacerecognition[edit] AutomaticfacedetectionwithOpenCV Whilehumanscanrecognizefaceswithoutmucheffort,[29]facialrecognitionisachallengingpatternrecognitionproblemincomputing.Facialrecognitionsystemsattempttoidentifyahumanface,whichisthree-dimensionalandchangesinappearancewithlightingandfacialexpression,basedonitstwo-dimensionalimage.Toaccomplishthiscomputationaltask,facialrecognitionsystemsperformfoursteps.Firstfacedetectionisusedtosegmentthefacefromtheimagebackground.Inthesecondstepthesegmentedfaceimageisalignedtoaccountforfacepose,imagesizeandphotographicproperties,suchasilluminationandgrayscale.Thepurposeofthealignmentprocessistoenabletheaccuratelocalizationoffacialfeaturesinthethirdstep,thefacialfeatureextraction.Featuressuchaseyes,noseandmoutharepinpointedandmeasuredintheimagetorepresenttheface.Thesoestablishedfeaturevectorofthefaceisthen,inthefourthstep,matchedagainstadatabaseoffaces.[30] Traditional[edit] SomeeigenfacesfromAT&TLaboratoriesCambridge Somefacerecognitionalgorithmsidentifyfacialfeaturesbyextractinglandmarks,orfeatures,fromanimageofthesubject'sface.Forexample,analgorithmmayanalyzetherelativeposition,size,and/orshapeoftheeyes,nose,cheekbones,andjaw.[31]Thesefeaturesarethenusedtosearchforotherimageswithmatchingfeatures.[32] Otheralgorithmsnormalizeagalleryoffaceimagesandthencompressthefacedata,onlysavingthedataintheimagethatisusefulforfacerecognition.Aprobeimageisthencomparedwiththefacedata.[33]Oneoftheearliestsuccessfulsystems[34]isbasedontemplatematchingtechniques[35]appliedtoasetofsalientfacialfeatures,providingasortofcompressedfacerepresentation. Recognitionalgorithmscanbedividedintotwomainapproaches:geometric,whichlooksatdistinguishingfeatures,orphoto-metric,whichisastatisticalapproachthatdistillsanimageintovaluesandcomparesthevalueswithtemplatestoeliminatevariances.Someclassifythesealgorithmsintotwobroadcategories:holisticandfeature-basedmodels.Theformerattemptstorecognizethefaceinitsentiretywhilethefeature-basedsubdivideintocomponentssuchasaccordingtofeaturesandanalyzeeachaswellasitsspatiallocationwithrespecttootherfeatures.[36] Popularrecognitionalgorithmsincludeprincipalcomponentanalysisusingeigenfaces,lineardiscriminantanalysis,elasticbunchgraphmatchingusingtheFisherfacealgorithm,thehiddenMarkovmodel,themultilinearsubspacelearningusingtensorrepresentation,andtheneuronalmotivateddynamiclinkmatching.[citationneeded][37]Modernfacialrecognitionsystemsmakeincreasinguseofmachinelearningtechniquessuchasdeeplearning.[38] Humanidentificationatadistance(HID)[edit] Toenablehumanidentificationatadistance(HID)low-resolutionimagesoffacesareenhancedusingfacehallucination.InCCTVimageryfacesareoftenverysmall.Butbecausefacialrecognitionalgorithmsthatidentifyandplotfacialfeaturesrequirehighresolutionimages,resolutionenhancementtechniqueshavebeendevelopedtoenablefacialrecognitionsystemstoworkwithimagerythathasbeencapturedinenvironmentswithahighsignal-to-noiseratio.Facehallucinationalgorithmsthatareappliedtoimagespriortothoseimagesbeingsubmittedtothefacialrecognitionsystemuseexample-basedmachinelearningwithpixelsubstitutionornearestneighbourdistributionindexesthatmayalsoincorporatedemographicandagerelatedfacialcharacteristics.Useoffacehallucinationtechniquesimprovestheperformanceofhighresolutionfacialrecognitionalgorithmsandmaybeusedtoovercometheinherentlimitationsofsuper-resolutionalgorithms.Facehallucinationtechniquesarealsousedtopre-treatimagerywherefacesaredisguised.Herethedisguise,suchassunglasses,isremovedandthefacehallucinationalgorithmisappliedtotheimage.Suchfacehallucinationalgorithmsneedtobetrainedonsimilarfaceimageswithandwithoutdisguise.Tofillintheareauncoveredbyremovingthedisguise,facehallucinationalgorithmsneedtocorrectlymaptheentirestateoftheface,whichmaybenotpossibleduetothemomentaryfacialexpressioncapturedinthelowresolutionimage.[39] 3-dimensionalrecognition[edit] 3Dmodelofahumanface Three-dimensionalfacerecognitiontechniqueuses3Dsensorstocaptureinformationabouttheshapeofaface.Thisinformationisthenusedtoidentifydistinctivefeaturesonthesurfaceofaface,suchasthecontouroftheeyesockets,nose,andchin.[40] Oneadvantageof3Dfacerecognitionisthatitisnotaffectedbychangesinlightinglikeothertechniques.Itcanalsoidentifyafacefromarangeofviewingangles,includingaprofileview.[40][32]Three-dimensionaldatapointsfromafacevastlyimprovetheprecisionoffacerecognition.3D-dimensionalfacerecognitionresearchisenabledbythedevelopmentofsophisticatedsensorsthatprojectstructuredlightontotheface.[41]3Dmatchingtechniquearesensitivetoexpressions,thereforeresearchersatTechnionappliedtoolsfrommetricgeometrytotreatexpressionsasisometries.[42]Anewmethodofcapturing3Dimagesoffacesusesthreetrackingcamerasthatpointatdifferentangles;onecamerawillbepointingatthefrontofthesubject,secondonetotheside,andthirdoneatanangle.Allthesecameraswillworktogethersoitcantrackasubject'sfaceinreal-timeandbeabletofacedetectandrecognize.[43] Thermalcameras[edit] Apseudocolorimageoftwopeopletakeninlong-wavelengthinfrared(body-temperaturethermal)light Adifferentformoftakinginputdataforfacerecognitionisbyusingthermalcameras,bythisprocedurethecameraswillonlydetecttheshapeoftheheadanditwillignorethesubjectaccessoriessuchasglasses,hats,ormakeup.[44]Unlikeconventionalcameras,thermalcamerascancapturefacialimageryeveninlow-lightandnighttimeconditionswithoutusingaflashandexposingthepositionofthecamera.[45]However,thedatabasesforfacerecognitionarelimited.Effortstobuilddatabasesofthermalfaceimagesdatebackto2004.[44]By2016severaldatabasesexisted,includingtheIIITD-PSEandtheNotreDamethermalfacedatabase.[46]Currentthermalfacerecognitionsystemsarenotabletoreliablydetectafaceinathermalimagethathasbeentakenofanoutdoorenvironment.[47] In2018,researchersfromtheU.S.ArmyResearchLaboratory(ARL)developedatechniquethatwouldallowthemtomatchfacialimageryobtainedusingathermalcamerawiththoseindatabasesthatwerecapturedusingaconventionalcamera.[48]Knownasacross-spectrumsynthesismethodduetohowitbridgesfacialrecognitionfromtwodifferentimagingmodalities,thismethodsynthesizeasingleimagebyanalyzingmultiplefacialregionsanddetails.[49]Itconsistsofanon-linearregressionmodelthatmapsaspecificthermalimageintoacorrespondingvisiblefacialimageandanoptimizationissuethatprojectsthelatentprojectionbackintotheimagespace.[45]ARLscientistshavenotedthattheapproachworksbycombiningglobalinformation(i.e.featuresacrosstheentireface)withlocalinformation(i.e.featuresregardingtheeyes,nose,andmouth).[50]AccordingtoperformancetestsconductedatARL,themulti-regioncross-spectrumsynthesismodeldemonstratedaperformanceimprovementofabout30%overbaselinemethodsandabout5%overstate-of-the-artmethods.[49] Application[edit] Socialmedia[edit] Foundedin2013,LookserywentontoraisemoneyforitsfacemodificationapponKickstarter.Aftersuccessfulcrowdfunding,LookserylaunchedinOctober2014.Theapplicationallowsvideochatwithothersthroughaspecialfilterforfacesthatmodifiesthelookofusers.Imageaugmentingapplicationsalreadyonthemarket,suchasFacetuneandPerfect365,werelimitedtostaticimages,whereasLookseryallowedaugmentedrealitytolivevideos.Inlate2015SnapChatpurchasedLooksery,whichwouldthenbecomeitslandmarklensesfunction.[51]Snapchatfilterapplicationsusefacedetectiontechnologyandonthebasisofthefacialfeaturesidentifiedinanimagea3Dmeshmaskislayeredovertheface.[52] DeepFaceisadeeplearningfacialrecognitionsystemcreatedbyaresearchgroupatFacebook.Itidentifieshumanfacesindigitalimages.Itemploysanine-layerneuralnetwithover120millionconnectionweights,andwastrainedonfourmillionimagesuploadedbyFacebookusers.[53][54]Thesystemissaidtobe97%accurate,comparedto85%fortheFBI'sNextGenerationIdentificationsystem.[55] TikTok'salgorithmhasbeenregardedasespeciallyeffective,butmanywerelefttowonderattheexactprogrammingthatcausedtheapptobesoeffectiveinguessingtheuser'sdesiredcontent.[56]InJune2020,Tiktokreleasedastatementregardingthe"ForYou"page,andhowtheyrecommendedvideostousers,whichdidnotincludefacialrecognition.[57]InFebruary2021,however,Tiktokagreedtoa$92 millionsettlementtoaUSlawsuitwhichallegedthattheapphadusedfacialrecognitioninbothuservideosanditsalgorithmtoidentifyage,genderandethnicity.[58] IDverification[edit] TheemerginguseoffacialrecognitionisintheuseofIDverificationservices.Manycompaniesandothersareworkinginthemarketnowtoprovidetheseservicestobanks,ICOs,andothere-businesses.[59]Facerecognitionhasbeenleveragedasaformofbiometricauthenticationforvariouscomputingplatformsanddevices;[32]Android4.0"IceCreamSandwich"addedfacialrecognitionusingasmartphone'sfrontcameraasameansofunlockingdevices,[60][61]whileMicrosoftintroducedfacerecognitionlogintoitsXbox360videogameconsolethroughitsKinectaccessory,[62]aswellasWindows10viaits"WindowsHello"platform(whichrequiresaninfrared-illuminatedcamera).[63]In2017,Apple'siPhoneXsmartphoneintroducedfacialrecognitiontotheproductlinewithits"FaceID"platform,whichusesaninfraredilluminationsystem.[64] FaceID[edit] AppleintroducedFaceIDontheflagshipiPhoneXasabiometricauthenticationsuccessortotheTouchID,afingerprintbasedsystem.FaceIDhasafacialrecognitionsensorthatconsistsoftwoparts:a"Romeo"modulethatprojectsmorethan30,000infrareddotsontotheuser'sface,anda"Juliet"modulethatreadsthepattern.[65]Thepatternissenttoalocal"SecureEnclave"inthedevice'scentralprocessingunit(CPU)toconfirmamatchwiththephoneowner'sface.[66] ThefacialpatternisnotaccessiblebyApple.Thesystemwillnotworkwitheyesclosed,inanefforttopreventunauthorizedaccess.[66]Thetechnologylearnsfromchangesinauser'sappearance,andthereforeworkswithhats,scarves,glasses,andmanysunglasses,beardandmakeup.[67]Italsoworksinthedark.Thisisdonebyusinga"FloodIlluminator",whichisadedicatedinfraredflashthatthrowsoutinvisibleinfraredlightontotheuser'sfacetoproperlyreadthe30,000facialpoints.[68] DeploymentofFRTforavailinggovernmentservices[edit] India[edit] Inaninterview,theNationalHealthAuthoritychiefDr.R.S.SharmasaidthatfacialrecognitiontechnologywouldbeusedinconjunctionwithAadhaartoauthenticatetheidentityofpeopleseekingvaccines.[69]Tenhumanrightsanddigitalrightsorganizationsandmorethan150individualssignedastatementbytheInternetFreedomFoundationthatraisedalarmagainstthedeploymentoffacialrecognitiontechnologyinthecentralgovernment'svaccinationdriveprocess.[70]Implementationofanerror-pronesystemwithoutadequatelegislationcontainingmandatorysafeguards,woulddeprivecitizensofessentialservicesandlinkingthisuntestedtechnologytothevaccinationroll-outinIndiawillonlyexcludepersonsfromthevaccinedeliverysystem.[71] InJuly,2021,apressreleasebytheGovernmentofMeghalayastatedthatfacialrecognitiontechnology(FRT)wouldbeusedtoverifytheidentityofpensionerstoissueaDigitalLifeCertificateusing“Pensioner’sLifeCertificationVerification”mobileapplication.[72]Thenotice,accordingtothepressrelease,purportstoofferpensioners“asecure,easyandhassle-freeinterfaceforverifyingtheirlivenesstothePensionDisbursingAuthoritiesfromthecomfortoftheirhomesusingsmartphones”.Mr.JadeJeremiahLyngdoh,alawstudent,sentalegalnoticetotherelevantauthoritieshighlightingthat“Theapplicationhasbeenrolledoutwithoutanyanchoringlegislationwhichgovernstheprocessingofpersonaldataandthus,lackslawfulnessandtheGovernmentisnotempoweredtoprocessdata.”[73] Deploymentinsecurityservices[edit] SwissEuropeansurveillance:facerecognitionandvehiclemake,model,colorandlicenseplatereader Commonwealth[edit] TheAustralianBorderForceandNewZealandCustomsServicehavesetupanautomatedborderprocessingsystemcalledSmartGatethatusesfacerecognition,whichcomparesthefaceofthetravellerwiththedatainthee-passportmicrochip.[74][75]AllCanadianinternationalairportsusefacialrecognitionaspartofthePrimaryInspectionKioskprogramthatcomparesatravelerfacetotheirphotostoredontheePassport.ThisprogramfirstcametoVancouverInternationalAirportinearly2017andwasrolleduptoallremaininginternationalairportsin2018–2019.[76] PoliceforcesintheUnitedKingdomhavebeentrialinglivefacialrecognitiontechnologyatpubliceventssince2015.[77]InMay2017,amanwasarrestedusinganautomaticfacialrecognition(AFR)systemmountedonavanoperatedbytheSouthWalesPolice.ArsTechnicareportedthat"thisappearstobethefirsttime[AFR]hasledtoanarrest".[78]However,a2018reportbyBigBrotherWatchfoundthatthesesystemswereupto98%inaccurate.[77]ThereportalsorevealedthattwoUKpoliceforces,SouthWalesPoliceandtheMetropolitanPolice,wereusinglivefacialrecognitionatpubliceventsandinpublicspaces.[79] InSeptember2019,SouthWalesPoliceuseoffacialrecognitionwasruledlawful.[79]Livefacialrecognitionhasbeentrialledsince2016inthestreetsofLondonandwillbeusedonaregularbasisfromMetropolitanPolicefrombeginningof2020.[80]InAugust2020theCourtofAppealruledthatthewaythefacialrecognitionsystemhadbeenusedbytheSouthWalesPolicein2017and2018violatedhumanrights.[81] UnitedStates[edit] Flightboardinggatewith"biometricfacescanners"developedbyU.S.CustomsandBorderProtectionatHartsfield–JacksonAtlantaInternationalAirport TheU.S.DepartmentofStateoperatesoneofthelargestfacerecognitionsystemsintheworldwithadatabaseof117millionAmericanadults,withphotostypicallydrawnfromdriver'slicensephotos.[82]Althoughitisstillfarfromcompletion,itisbeingputtouseincertaincitiestogivecluesastowhowasinthephoto.TheFBIusesthephotosasaninvestigativetool,notforpositiveidentification.[83]Asof2016,facialrecognitionwasbeingusedtoidentifypeopleinphotostakenbypoliceinSanDiegoandLosAngeles(notonreal-timevideo,andonlyagainstbookingphotos)[84]andusewasplannedinWestVirginiaandDallas.[85] InrecentyearsMarylandhasusedfacerecognitionbycomparingpeople'sfacestotheirdriver'slicensephotos.ThesystemdrewcontroversywhenitwasusedinBaltimoretoarrestunrulyprotestersafterthedeathofFreddieGrayinpolicecustody.[86]Manyotherstatesareusingordevelopingasimilarsystemhoweversomestateshavelawsprohibitingitsuse. TheFBIhasalsoinstituteditsNextGenerationIdentificationprogramtoincludefacerecognition,aswellasmoretraditionalbiometricslikefingerprintsandirisscans,whichcanpullfrombothcriminalandcivildatabases.[87]ThefederalGeneralAccountabilityOfficecriticizedtheFBIfornotaddressingvariousconcernsrelatedtoprivacyandaccuracy.[88] Startingin2018,U.S.CustomsandBorderProtectiondeployed"biometricfacescanners"atU.S.airports.Passengerstakingoutboundinternationalflightscancompletethecheck-in,securityandtheboardingprocessaftergettingfacialimagescapturedandverifiedbymatchingtheirIDphotosstoredonCBP'sdatabase.ImagescapturedfortravelerswithU.S.citizenshipwillbedeletedwithinupto12-hours.TSAhadexpresseditsintentiontoadoptasimilarprogramfordomesticairtravelduringthesecuritycheckprocessinthefuture.TheAmericanCivilLibertiesUnionisoneoftheorganizationsagainsttheprogram,concerningthattheprogramwillbeusedforsurveillancepurposes.[89] In2019,researchersreportedthatImmigrationandCustomsEnforcementusesfacialrecognitionsoftwareagainststatedriver'slicensedatabases,includingforsomestatesthatprovidelicensestoundocumentedimmigrants.[88] China[edit] In2006,theSkynetProjectwasinitiatedbytheChineseGovernmenttoimplementCCTVsurveillancenationwideandasof2018,therehasbeen20millioncameras,manyofwhichcapableofreal-timefacialrecognition,deployedacrossthecountryforthisproject[90]SomeofficialclaimthatthecurrentSkynetsystemcanscantheentireChinesepopulationinonesecondandtheworldpopulationintwoseconds.[91] BoardinggateswithfacialrecognitiontechnologyatBeijingWestrailwaystation In2017,theQingdaopolicewasabletoidentifytwenty-fivewantedsuspectsusingfacialrecognitionequipmentattheQingdaoInternationalBeerFestival,oneofwhichhadbeenontherunfor10years.[92]Theequipmentworksbyrecordinga15-secondvideoclipandtakingmultiplesnapshotsofthesubject.Thatdataiscomparedandanalyzedwithimagesfromthepolicedepartment'sdatabaseandwithin20minutes,thesubjectcanbeidentifiedwitha98.1%accuracy.[93] In2018,ChinesepoliceinZhengzhouandBeijingwereusingsmartglassestotakephotoswhicharecomparedagainstagovernmentdatabaseusingfacialrecognitiontoidentifysuspects,retrieveanaddress,andtrackpeoplemovingbeyondtheirhomeareas.[94][95] Asoflate2017,ChinahasdeployedfacialrecognitionandartificialintelligencetechnologyinXinjiang.Reportersvisitingtheregionfoundsurveillancecamerasinstalledeveryhundredmetersorsoinseveralcities,aswellasfacialrecognitioncheckpointsatareaslikegasstations,shoppingcenters,andmosqueentrances.[96][97]InMay2019,HumanRightsWatchreportedfindingFace++codeintheIntegratedJointOperationsPlatform(IJOP),apolicesurveillanceappusedtocollectdataon,andtracktheUighurcommunityinXinjiang.[98]HumanRightsWatchreleasedacorrectiontoitsreportinJune2019statingthattheChinesecompanyMegviididnotappeartohavecollaboratedonIJOP,andthattheFace++codeintheappwasinoperable.[99]InFebruary2020,followingtheCoronavirusoutbreak,MegviiappliedforabankloantooptimizethebodytemperaturescreeningsystemithadlaunchedtohelpidentifypeoplewithsymptomsofaCoronavirusinfectionincrowds.IntheloanapplicationMegviistatedthatitneededtoimprovetheaccuracyofidentifyingmaskedindividuals.[100] ManypublicplacesinChinaareimplementedwithfacialrecognitionequipment,includingrailwaystations,airports,touristattractions,expos,andofficebuildings.InOctober2019,aprofessoratZhejiangSci-TechUniversitysuedtheHangzhouSafariParkforabusingprivatebiometricinformationofcustomers.ThesafariparkusesfacialrecognitiontechnologytoverifytheidentitiesofitsYearCardholders.Anestimated300touristsitesinChinahaveinstalledfacialrecognitionsystemsandusethemtoadmitvisitors.ThiscaseisreportedtobethefirstontheuseoffacialrecognitionsystemsinChina.[101]InAugust2020,RadioFreeAsiareportedthatin2019GengGuanjun,acitizenofTaiyuanCitywhohadusedtheWeChatappbyTencenttoforwardavideotoafriendintheUnitedStateswassubsequentlyconvictedonthechargeofthecrime"pickingquarrelsandprovokingtroubles".TheCourtdocumentsshowedthattheChinesepoliceusedafacialrecognitionsystemtoidentifyGengGuanjunasan"overseasdemocracyactivist"andthatChina'snetworkmanagementandpropagandadepartmentsdirectlymonitorWeChatusers.[102] In2019,ProtestorsinHongKongdestroyedsmartlamppostsamidconcernstheycouldcontaincamerasandfacialrecognitionsystemusedforsurveillancebyChineseauthorities.[103] India[edit] Eventhoughfacialrecognitiontechnology(FRT)isnotfullyaccurate,[104]itisbeingincreasinglydeployedforidentificationpurposesbythepoliceinIndia.FRTsystemsgenerateaprobabilitymatchscore,oraconfidencescorebetweenthesuspectwhoistobeidentifiedandthedatabaseofidentifiedcriminalsthatisavailablewiththepolice.TheNationalAutomatedFacialRecognitionSystem(AFRS)[105]isalreadybeingdevelopedbytheNationalCrimeRecordsBureau(NCRB),abodyconstitutedundertheMinistryofHomeAffairs.Theprojectseekstodevelopanddeployanationaldatabaseofphotographswhichwouldcomportwithafacialrecognitiontechnologysystembythecentralandstatesecurityagencies.TheInternetFreedomFoundationhasflaggedconcernsregardingtheproject.[106]TheNGOhashighlightedthattheaccuracyofFRTsystemsare"routinelyexaggeratedandtherealnumbersleavemuchtobedesired.[106]TheimplementationofsuchfaultyFRTsystemswouldleadtohighratesoffalsepositivesandfalsenegativesinthisrecognitionprocess.”  UndertheSupremeCourtofIndia'sdecisionin[[JusticeK.S.PuttaswamyvsUnionofIndia(2201710SCC1),anyjustifiableintrusionbytheStateintopeople'srighttoprivacy,whichisprotectedasafundamentalrightunderArticle21oftheConstitution,mustconfirmtocertainthresholds,namely:legality,necessity,proportionalityandproceduralsafeguards.[107]AspertheInternetFreedomFoundation,theNationalAutomatedFacialRecognitionSystem(AFRS)proposalfailstomeetanyofthesethresholds,citing"absenceoflegality,""manifestarbitrariness,"and"absenceofsafeguardsandaccountability."[108] WhilethenationallevelAFRSprojectisstillintheworks,policedepartmentsinvariousstatesinIndiaarealreadydeployingfacialrecognitiontechnologysystems,suchas:TSCOP+CCTNSinTelangana,[109]PunjabArtificialIntelligenceSystem(PAIS)inPunjab,[110]TrinetrainUttarPradesh,[111]PoliceArtificialIntelligenceSysteminUttarakhand,[112]AFRSinDelhi,AutomatedMultimodalBiometricIdentificationSystem(AMBIS)inMaharashtra,FaceTagrinTamilNadu.TheCrimeandCriminalTrackingNetworkandSystems(CCTNS),whichisaMissionModeProjectundertheNationale-GovernancePlan(NeGP),[113]isviewedasasystemwhichwouldconnectpolicestationsacrossIndia,andhelpthem"talk"[114]toeachother.Theproject'sobjectiveistodigitizeallFIR-relatedinformation,includingFIRsregistered,aswellascasesinvestigated,chargesheetsfiled,andsuspectsandwantedpersonsinallpolicestations.ThisshallconstituteanationaldatabaseofcrimeandcriminalsinIndia.CCTNSisbeingimplementedwithoutadataprotectionlawinplace.CCTNSisproposedtobeintegratedwiththeAFRS,arepositoryofallcrimeandcriminalrelatedfacialdatawhichcanbedeployedtopurportedlyidentifyorverifyapersonfromavarietyofinputsrangingfromimagestovideos.[115]Thishasraisedprivacyconcernsfromcivilsocietyorganizationsandprivacyexperts.Boththeprojectshavebeencensuredasinstrumentsof"masssurveillance"atthehandsofthestate.[116]InRajasthan,'RajCop,'apoliceapphasbeenrecentlyintegratedwithafacialrecognitionmodulewhichcanmatchthefaceofasuspectagainstadatabaseofknownpersonsinreal-time.RajasthanpoliceisincurrentlyworkingtowidentheambitofthismodulebymakingitmandatorytouploadphotographsofallarrestedpersonsinCCTNSdatabase,whichwill"helpdeveloparichdatabaseofknownoffenders."[117] HelmetsfixedwithcamerahavebeendesignedandbeingusedbyRajasthanpoliceinlawandordersituationstocapturepoliceactionandactivitiesof“themiscreants,whichcanlaterserveasevidenceduringtheinvestigationofsuchcases.”[117]PAIS(PunjabArtificialIntelligenceSystem),Appemploysdeeplearning,machinelearning,andfacerecognitionfortheidentificationofcriminalstoassistpolicepersonnel.[117]ThestateofTelanganahasinstalled8lakhCCTVcameras,[117]withitscapitalcityHyderabadslowlyturningintoasurveillancecapital.[118] Afalsepositivehappenswhenfacialrecognitiontechnologymisidentifiesapersontobesomeonetheyarenot,thatis,ityieldsanincorrectpositiveresult.Theyoftenresultsindiscriminationandstrengtheningofexistingbiases.Forexample,in2018,DelhiPolicereportedthatitsFRTsystemhadanaccuracyrateof2%,whichsankto1%in2019.TheFRTsystemevenfailedtodistinguishaccuratelybetweendifferentsexes.[119] ThegovernmentofDelhiincollaborationwithIndianSpaceResearchOrganisation(ISRO)isdevelopinganewtechnologycalledCrimeMappingAnalyticsandPredictiveSystem(CMAPS).Theprojectaimstodeployspacetechnologyfor"controllingcrimeandmaintaininglawandorder."[117]Thesystemwillbeconnectedtoadatabasecontainingdataofcriminals.[117]Thetechnologyisenvisagedtobedeployedtocollectreal-timedataatthecrimescene.[117] InareplydatedNovember25,2020toaRighttoInformationrequestfiledbytheInternetFreedomFoundationseekinginformationaboutthefacialrecognitionsystembeingusedbytheDelhiPolice(withreferencenumberDEPOL/R/E/20/07128),[120]theOfficeoftheDeputyCommissionerofPolicecumPublicInformationOfficer:Crimestatedthattheycannotprovidetheinformationundersection8(d)oftheRighttoInformationAct,2005.[121] ARighttoInformation(RTI)requestdatedJuly30,2020wasfiledwiththeOfficeoftheCommissioner,KolkataPolice,seekinginformationaboutthefacialrecognitiontechnologythatthedepartmentwasusing.[122]Theinformationsoughtwasdenied[123]statingthatthedepartmentwasexemptedfromdisclosureundersection24(4)oftheRTIAct. LatinAmerica[edit] Inthe2000Mexicanpresidentialelection,theMexicangovernmentemployedfacerecognitionsoftwaretopreventvoterfraud.Someindividualshadbeenregisteringtovoteunderseveraldifferentnames,inanattempttoplacemultiplevotes.Bycomparingnewfaceimagestothosealreadyinthevoterdatabase,authoritieswereabletoreduceduplicateregistrations.[124] InColombiapublictransportbussesarefittedwithafacialrecognitionsystembyFaceFirstInctoidentifypassengersthataresoughtbytheNationalPoliceofColombia.FaceFirstIncalsobuiltthefacialrecognitionsystemforTocumenInternationalAirportinPanama.ThefacerecognitionsystemisdeployedtoidentifyindividualsamongthetravelersthataresoughtbythePanamanianNationalPoliceorInterpol.[125]TocumenInternationalAirportoperatesanairport-widesurveillancesystemusinghundredsoflivefacerecognitioncamerastoidentifywantedindividualspassingthroughtheairport.ThefacerecognitionsystemwasinitiallyinstalledaspartofaUS$11 millioncontractandincludedacomputerclusterofsixtycomputers,afiber-opticcablenetworkfortheairportbuildings,aswellastheinstallationof150surveillancecamerasintheairportterminalandatabout30airportgates.[126] Atthe2014FIFAWorldCupinBraziltheFederalPoliceofBrazilusedfacerecognitiongoggles.Facerecognitionsystems"madeinChina"werealsodeployedatthe2016SummerOlympicsinRiodeJaneiro.[125]NuctechCompanyprovided145insepctionterminalsforMaracanãStadiumand55terminalsfortheDeodoroOlympicPark.[127] EuropeanUnion[edit] Policeforcesinatleast21countriesoftheEuropeanUnionuse,orplantouse,facialrecognitionsystems,eitherforadministrativeorcriminalpurposes.[128] Greece[edit] GreekpolicepassedacontractwithIntracom-Telecomfortheprovisionofatleast1,000devicesequippedwithlivefacialrecognitionsystem.Thedeliveryisexpectedbeforethesummer2021.Thetotalvalueofthecontractisover4millioneuros,paidforinlargepartbytheInternalSecurityFundoftheEuropeanCommission.[129] Italy[edit] Italianpoliceacquiredafacerecognitionsystemin2017,SistemaAutomaticoRiconoscimentoImmagini(SARI).InNovember2020,theInteriorministryannouncedplanstouseitinreal-timetoidentifypeoplesuspectedofseekingasylum.[130] TheNetherlands[edit] TheNetherlandshasdeployedfacialrecognitionandartificialintelligencetechnologysince2016.[131]ThedatabaseoftheDutchpolicecurrentlycontainsover2.2 millionpicturesof1.3 millionDutchcitizens.Thisaccountsforabout8%ofthepopulation.InTheNetherlands,facerecognitionisnotusedbythepoliceonmunicipalCCTV.[132] SouthAfrica[edit] InSouthAfrica,in2016,thecityofJohannesburgannounceditwasrollingoutsmartCCTVcamerascompletewithautomaticnumberplaterecognitionandfacialrecognition.[133] Deploymentinretailstores[edit] TheUSfirm3VR,nowIdentiv,isanexampleofavendorwhichbeganofferingfacialrecognitionsystemsandservicestoretailersasearlyas2007.[134]In2012,thecompanyadvertisedbenefitssuchas"dwellandqueuelineanalyticstodecreasecustomerwaittimes","facialsurveillanceanalytic[s]tofacilitatepersonalizedcustomergreetingsbyemployees"andtheabilityto"[c]reateloyaltyprogramsbycombiningPointofsale(POS)datawithfacialrecognition".[135] UnitedStates[edit] In2018,theNationalRetailFederationLossPreventionResearchCouncilcalledfacialrecognitiontechnology"apromisingnewtool"worthevaluating.[136] InJuly2020,theReutersnewsagencyreportedthatduringthe2010sthepharmacychainRiteAidhaddeployedfacialrecognitionvideosurveillancesystemsandcomponentsfromFaceFirst,DeepCamLLC,andothervendorsatsomeretaillocationsintheUnitedStates.[136]CathyLangley,RiteAid'svicepresidentofassetprotection,usedthephrase"featurematching"torefertothesystemsandsaidthatusageofthesystemsresultedinlessviolenceandorganizedcrimeinthecompany'sstores,whileformervicepresidentofassetprotectionBobOberosleremphasizedimprovedsafetyforstaffandareducedneedfortheinvolvementoflawenforcementorganizations.[136]Ina2020statementtoReutersinresponsetothereporting,RiteAidsaidthatithadceasedusingthefacialrecognitionsoftwareandswitchedoffthecameras.[136] AccordingtodirectorReadHayesoftheNationalRetailFederationLossPreventionResearchCouncil,RiteAid'ssurveillanceprogramwaseitherthelargestoroneofthelargestprogramsinretail.[136]TheHomeDepot,Menards,Walmart,and7-ElevenareamongotherUSretailersalsoengagedinlarge-scalepilotprogramsordeploymentsoffacialrecognitiontechnology.[136] OftheRiteAidstoresexaminedbyReutersin2020,thoseincommunitieswherepeopleofcolormadeupthelargestracialorethnicgroupwerethreetimesaslikelytohavethetechnologyinstalled,[136]raisingconcernsrelatedtothesubstantialhistoryofracialsegregationandracialprofilingintheUnitedStates.RiteAidsaidthattheselectionoflocationswas"data-driven",basedonthethefthistoriesofindividualstores,localandnationalcrimedata,andsiteinfrastructure.[136] Australia[edit] In2019,facialrecognitiontopreventtheftwasinuseatSydney'sStarCasinoandwasalsodeployedatgamingvenuesinNewZealand.[137] InJune2022,consumergroupCHOICEreportedfacialrecognitionwasinuseinAustraliaatKmart,Bunnings,andTheGoodGuys.TheGoodGuyssubsequentlysuspendedthetechnologypendingalegalchallengebyCHOICEtotheOfficeoftheAustralianInformationCommissioner,whileBunningskeptthetechnologyinuseandKmartmaintaineditstrialofthetechnology.[138] Additionaluses[edit] Disney'sMagicKingdom,nearOrlando,Florida,duringatrialofafacialrecognitiontechnologyforparkentry AttheAmericanfootballchampionshipgameSuperBowlXXXVinJanuary2001,policeinTampaBay,FloridausedViisagefacerecognitionsoftwaretosearchforpotentialcriminalsandterroristsinattendanceattheevent.19peoplewithminorcriminalrecordswerepotentiallyidentified.[139][140] Facerecognitionsystemshavealsobeenusedbyphotomanagementsoftwaretoidentifythesubjectsofphotographs,enablingfeaturessuchassearchingimagesbyperson,aswellassuggestingphotostobesharedwithaspecificcontactiftheirpresenceweredetectedinaphoto.[141][142]By2008facialrecognitionsystemsweretypicallyusedasaccesscontrolinsecuritysystems.[143] TheUnitedStates'popularmusicandcountrymusiccelebrityTaylorSwiftsurreptitiouslyemployedfacialrecognitiontechnologyataconcertin2018.Thecamerawasembeddedinakiosknearaticketboothandscannedconcert-goersastheyenteredthefacilityforknownstalkers.[144] OnAugust18,2019,TheTimesreportedthattheUAE-ownedManchesterCityhiredaTexas-basedfirm,BlinkIdentity,todeployfacialrecognitionsystemsinadriverprogram.Theclubhasplannedasinglesuper-fastlaneforthesupportersattheEtihadstadium.[145]However,civilrightsgroupscautionedtheclubagainsttheintroductionofthistechnology,sayingthatitwouldrisk"normalisingamasssurveillancetool".ThepolicyandcampaignsofficeratLiberty,HannahCouchmansaidthatManCity'smoveisalarming,sincethefanswillbeobligedtosharedeeplysensitivepersonalinformationwithaprivatecompany,wheretheycouldbetrackedandmonitoredintheireverydaylives.[146] In2019,casinosinAustraliaandNewZealandrolledoutfacialrecognitiontopreventtheft,andarepresentativeofSydney'sStarCasinosaidtheywouldalsoprovide'customerservice'likewelcomingapatronbacktoabar.[137] InAugust2020,amidtheCOVID-19pandemicintheUnitedStates,AmericanfootballstadiumsofNewYorkandLosAngelesannouncedtheinstallationoffacialrecognitionforupcomingmatches.Thepurposeistomaketheentryprocessastouchlessaspossible.[147]Disney'sMagicKingdom,nearOrlando,Florida,likewiseannouncedatestoffacialrecognitiontechnologytocreateatouchlessexperienceduringthepandemic;thetestwasoriginallyslatedtotakeplacebetweenMarch23andApril23,2021,butthelimitedtimeframehadbeenremovedasoflateApril.[148] Advantagesanddisadvantages[edit] Comparedtootherbiometricsystems[edit] In2006,theperformanceofthelatestfacerecognitionalgorithmswasevaluatedintheFaceRecognitionGrandChallenge(FRGC).High-resolutionfaceimages,3-Dfacescans,andirisimageswereusedinthetests.Theresultsindicatedthatthenewalgorithmsare10timesmoreaccuratethanthefacerecognitionalgorithmsof2002and100timesmoreaccuratethanthoseof1995.Someofthealgorithmswereabletooutperformhumanparticipantsinrecognizingfacesandcoulduniquelyidentifyidenticaltwins.[40][149] Onekeyadvantageofafacialrecognitionsystemthatitisabletoperformmassidentificationasitdoesnotrequirethecooperationofthetestsubjecttowork.Properlydesignedsystemsinstalledinairports,multiplexes,andotherpublicplacescanidentifyindividualsamongthecrowd,withoutpassers-byevenbeingawareofthesystem.[150]However,ascomparedtootherbiometrictechniques,facerecognitionmaynotbemostreliableandefficient.Qualitymeasuresareveryimportantinfacialrecognitionsystemsaslargedegreesofvariationsarepossibleinfaceimages.Factorssuchasillumination,expression,poseandnoiseduringfacecapturecanaffecttheperformanceoffacialrecognitionsystems.[150]Amongallbiometricsystems,facialrecognitionhasthehighestfalseacceptanceandrejectionrates,[150]thusquestionshavebeenraisedontheeffectivenessoffacerecognitionsoftwareincasesofrailwayandairportsecurity.[151] Weaknesses[edit] RalphGross,aresearcherattheCarnegieMellonRoboticsInstitutein2008,describesoneobstaclerelatedtotheviewingangleoftheface:"Facerecognitionhasbeengettingprettygoodatfullfrontalfacesand20degreesoff,butassoonasyougotowardsprofile,there'vebeenproblems."[40]Besidestheposevariations,low-resolutionfaceimagesarealsoveryhardtorecognize.Thisisoneofthemainobstaclesoffacerecognitioninsurveillancesystems.[152] Facerecognitionislesseffectiveiffacialexpressionsvary.Abigsmilecanrenderthesystemlesseffective.Forinstance:Canada,in2009,allowedonlyneutralfacialexpressionsinpassportphotos.[153] Thereisalsoinconstancyinthedatasetsusedbyresearchers.Researchersmayuseanywherefromseveralsubjectstoscoresofsubjectsandafewhundredimagestothousandsofimages.Itisimportantforresearcherstomakeavailablethedatasetstheyusedtoeachother,orhaveatleastastandarddataset.[154] Facialrecognitionsystemshavebeencriticizedforupholdingandjudgingbasedonabinarygenderassumption.[155][156][157][158][159][160][161][162][163]Whenclassifyingthefacesofcisgenderindividualsintomaleorfemale,thesesystemsareoftenveryaccurate,[155]howeverweretypicallyconfusedorunabletodeterminethegenderidentityoftransgenderandnon-binarypeople.[155]Gendernormsarebeingupheldbythesesystems,somuchsothatevenwhenshownaphotoofacisgendermalewithlonghair,algorithmswassplitbetweenfollowingthegendernormofmaleshavingshorthair,andthemasculinefacialfeaturesandbecameconfused.[155][163]Thisaccidentalmisgenderingofpeoplecanbeveryharmfulforthosewhodonotidentifywiththeirsexassignedatbirth,bydisregardingandinvalidatingtheirgenderidentity.Thisisalsoharmfulforpeoplewhodonotascribetotraditionalgendernorms,becauseitinvalidatestheirgenderexpression,regardlessoftheirgenderidentity. Ineffectiveness[edit] CriticsofthetechnologycomplainthattheLondonBoroughofNewhamschemehas,asof2004[update],neverrecognizedasinglecriminal,despiteseveralcriminalsinthesystem'sdatabaselivingintheBoroughandthesystemhasbeenrunningforseveralyears."Notonce,asfarasthepoliceknow,hasNewham'sautomaticfacerecognitionsystemspottedalivetarget."[140][164]Thisinformationseemstoconflictwithclaimsthatthesystemwascreditedwitha34%reductionincrime(hencewhyitwasrolledouttoBirminghamalso).[165] Anexperimentin2002bythelocalpolicedepartmentinTampa,Florida,hadsimilarlydisappointingresults.[140]AsystematBoston'sLoganAirportwasshutdownin2003afterfailingtomakeanymatchesduringatwo-yeartestperiod.[166] In2014,Facebookstatedthatinastandardizedtwo-optionfacialrecognitiontest,itsonlinesystemscored97.25%accuracy,comparedtothehumanbenchmarkof97.5%.[167] Systemsareoftenadvertisedashavingaccuracynear100%;thisismisleadingasthestudiesoftenusemuchsmallersamplesizesthanwouldbenecessaryforlargescaleapplications.Becausefacialrecognitionisnotcompletelyaccurate,itcreatesalistofpotentialmatches.Ahumanoperatormustthenlookthroughthesepotentialmatchesandstudiesshowtheoperatorspickthecorrectmatchoutofthelistonlyabouthalfthetime.Thiscausestheissueoftargetingthewrongsuspect.[83][168] Controversies[edit] Privacyviolations[edit] CivilrightsorganizationsandprivacycampaignerssuchastheElectronicFrontierFoundation,BigBrotherWatchandtheACLUexpressconcernthatprivacyisbeingcompromisedbytheuseofsurveillancetechnologies.[169][77][170]Facerecognitioncanbeusednotjusttoidentifyanindividual,butalsotounearthotherpersonaldataassociatedwithanindividual–suchasotherphotosfeaturingtheindividual,blogposts,socialmediaprofiles,Internetbehavior,andtravelpatterns.[171]Concernshavebeenraisedoverwhowouldhaveaccesstotheknowledgeofone'swhereaboutsandpeoplewiththematanygiventime.[172]Moreover,individualshavelimitedabilitytoavoidorthwartfacerecognitiontrackingunlesstheyhidetheirfaces.Thisfundamentallychangesthedynamicofday-to-dayprivacybyenablinganymarketer,governmentagency,orrandomstrangertosecretlycollecttheidentitiesandassociatedpersonalinformationofanyindividualcapturedbythefacerecognitionsystem.[171]Consumersmaynotunderstandorbeawareofwhattheirdataisbeingusedfor,whichdeniesthemtheabilitytoconsenttohowtheirpersonalinformationgetsshared.[172] InJuly2015,theUnitedStatesGovernmentAccountabilityOfficeconductedaReporttotheRankingMember,SubcommitteeonPrivacy,TechnologyandtheLaw,CommitteeontheJudiciary,U.S.Senate.Thereportdiscussedfacialrecognitiontechnology'scommercialuses,privacyissues,andtheapplicablefederallaw.Itstatesthatpreviously,issuesconcerningfacialrecognitiontechnologywerediscussedandrepresenttheneedforupdatingtheprivacylawsoftheUnitedStatessothatfederallawcontinuallymatchestheimpactofadvancedtechnologies.Thereportnotedthatsomeindustry,government,andprivateorganizationswereintheprocessofdeveloping,orhavedeveloped,"voluntaryprivacyguidelines".Theseguidelinesvariedbetweenthestakeholders,buttheiroverallaimwastogainconsentandinformcitizensoftheintendeduseoffacialrecognitiontechnology.Accordingtothereportthevoluntaryprivacyguidelineshelpedtocounteracttheprivacyconcernsthatarisewhencitizensareunawareofhowtheirpersonaldatagetsputtouse.[172] In2016,RussiancompanyNtechLabcausedaprivacyscandalintheinternationalmediawhenitlaunchedtheFindFacefacerecognitionsystemwiththepromisethatRussianuserscouldtakephotosofstrangersinthestreetandlinkthemtoasocialmediaprofileonthesocialmediaplatformVkontakte(VT).[173]InDecember2017,FacebookrolledoutanewfeaturethatnotifiesauserwhensomeoneuploadsaphotothatincludeswhatFacebookthinksistheirface,eveniftheyarenottagged.Facebookhasattemptedtoframethenewfunctionalityinapositivelight,amidstpriorbacklashes.[174]Facebook'sheadofprivacy,RobSherman,addressedthisnewfeatureasonethatgivespeoplemorecontrolovertheirphotosonline."We'vethoughtaboutthisasareallyempoweringfeature,"hesays."Theremaybephotosthatexistthatyoudon'tknowabout."[175]Facebook'sDeepFacehasbecomethesubjectofseveralclassactionlawsuitsundertheBiometricInformationPrivacyAct,withclaimsallegingthatFacebookiscollectingandstoringfacerecognitiondataofitsuserswithoutobtaininginformedconsent,indirectviolationofthe2008BiometricInformationPrivacyAct(BIPA).[176]ThemostrecentcasewasdismissedinJanuary2016becausethecourtlackedjurisdiction.[177]IntheUS,surveillancecompaniessuchasClearviewAIarerelyingontheFirstAmendmenttotheUnitedStatesConstitutiontodatascrapeuseraccountsonsocialmediaplatformsfordatathatcanbeusedinthedevelopmentoffacialrecognitionsystems.[178] In2019,theFinancialTimesfirstreportedthatfacialrecognitionsoftwarewasinuseintheKing'sCrossareaofLondon.[179]ThedevelopmentaroundLondon'sKing'sCrossmainlinestationincludesshops,offices,Google'sUKHQandpartofStMartin'sCollege.AccordingtotheUKInformationCommissioner'sOffice:"Scanningpeople'sfacesastheylawfullygoabouttheirdailylives,inordertoidentifythem,isapotentialthreattoprivacythatshouldconcernusall."[180][181]TheUKInformationCommissionerElizabethDenhamlaunchedaninvestigationintotheuseoftheKing'sCrossfacialrecognitionsystem,operatedbythecompanyArgent.InSeptember2019itwasannouncedbyArgentthatfacialrecognitionsoftwarewouldnolongerbeusedatKing'sCross.ArgentclaimedthatthesoftwarehadbeendeployedbetweenMay2016andMarch2018ontwocamerascoveringapedestrianstreetrunningthroughthecentreofthedevelopment.[182]InOctober2019,areportbythedeputyLondonmayorSophieLindenrevealedthatinasecretdealtheMetropolitanPolicehadpassedphotosofsevenpeopletoArgentforuseintheirKing'scrossfacialrecognitionsystem.[183] AutomatedFacialRecognitionwastrialledbytheSouthWalesPoliceonmultipleoccasionsbetween2017and2019.Theuseofthetechnologywaschallengedincourtbyaprivateindividual,EdwardBridges,withsupportfromthecharityLiberty(caseknownasR(Bridges)vChiefConstableSouthWalesPolice).ThecasewasheardintheCourtofAppealandajudgementwasgiveninAugust2020.[184]ThecasearguedthattheuseofFacialRecognitionwasaprivacyviolationonthebasisthattherewasinsufficientlegalframeworkorproportionalityintheuseofFacialRecognitionandthatitsusewasinviolationoftheDataProtectionActs1998and2018.ThecasewasdecidedinfavourofBridgesanddidnotawarddamages.Thecasewassettledviaadeclarationofwrongdoing.[184]Inresponsetothecase,theBritishGovernmenthasrepeatedlyattemptedtopassaBillregulatingtheuseofFacialRecognitioninpublicspaces.TheproposedBillshaveattemptedtoappointaCommissionerwiththeabilitytoregulateFacialRecognitionusebyGovernmentServicesinasimilarmannertotheCommissionerforCCTV.SuchaBillhasyettocomeintoforce[correctasofSeptember2021].[110] Imperfecttechnologyinlawenforcement[edit] Asof2018[update],itisstillcontestedastowhetherornotfacialrecognitiontechnologyworkslessaccuratelyonpeopleofcolor.[185]OnestudybyJoyBuolamwini(MITMediaLab)andTimnitGebru(MicrosoftResearch)foundthattheerrorrateforgenderrecognitionforwomenofcolorwithinthreecommercialfacialrecognitionsystemsrangedfrom23.8%to36%,whereasforlighter-skinnedmenitwasbetween0.0and1.6%.Overallaccuracyratesforidentifyingmen(91.9%)werehigherthanforwomen(79.4%),andnoneofthesystemsaccommodatedanon-binaryunderstandingofgender.[186]Italsoshowedthatthedatasetsusedtotraincommercialfacialrecognitionmodelswereunrepresentativeofthebroaderpopulationandskewedtowardlighter-skinnedmales.However,anotherstudyshowedthatseveralcommercialfacialrecognitionsoftwaresoldtolawenforcementofficesaroundthecountryhadalowerfalsenon-matchrateforblackpeoplethanforwhitepeople.[187] Expertsfearthatfacerecognitionsystemsmayactuallybehurtingcitizensthepoliceclaimstheyaretryingtoprotect.[188]Itisconsideredanimperfectbiometric,andinastudyconductedbyGeorgetownUniversityresearcherClareGarvie,sheconcludedthat"there'snoconsensusinthescientificcommunitythatitprovidesapositiveidentificationofsomebody."[189]Itisbelievedthatwithsuchlargemarginsoferrorinthistechnology,bothlegaladvocatesandfacialrecognitionsoftwarecompaniessaythatthetechnologyshouldonlysupplyaportionofthecase–noevidencethatcanleadtoanarrestofanindividual.[189]Thelackofregulationsholdingfacialrecognitiontechnologycompaniestorequirementsofraciallybiasedtestingcanbeasignificantflawintheadoptionofuseinlawenforcement.CyberExtruder,acompanythatmarketsitselftolawenforcementsaidthattheyhadnotperformedtestingorresearchonbiasintheirsoftware.CyberExtruderdidnotethatsomeskincolorsaremoredifficultforthesoftwaretorecognizewithcurrentlimitationsofthetechnology."Justasindividualswithverydarkskinarehardtoidentifywithhighsignificanceviafacialrecognition,individualswithverypaleskinarethesame,"saidBlakeSenftner,aseniorsoftwareengineeratCyberExtruder.[189] TheUnitedState'sNationalInstituteofStandardsandTechnology(NIST)carriedoutextensivetestingofFRTsystem1:1verification[190]and1:manyidentification.[190]ItalsotestedforthedifferingaccuracyofFRTacrossdifferentdemographicgroups.Theindependentstudyconcludedatpresent,noFRTsystemhas100%accuracy.[191] Dataprotection[edit] In2010,PerupassedtheLawforPersonalDataProtection,whichdefinesbiometricinformationthatcanbeusedtoidentifyanindividualassensitivedata.In2012,ColombiapassedacomprehensiveDataProtectionLawwhichdefinesbiometricdataassenstiviteinformation.[125]AccordingtoArticle9(1)oftheEU's2016GeneralDataProtectionRegulation(GDPR)theprocessingofbiometricdataforthepurposeof"uniquelyidentifyinganaturalperson"issensitiveandthefacialrecognitiondataprocessedinthiswaybecomessensitivepersonaldata.InresponsetotheGDPRpassingintothelawofEUmemberstates,EUbasedresearchersvoicedconcernthatiftheywererequiredundertheGDPRtoobtainindividual'sconsentfortheprocessingoftheirfacialrecognitiondata,afacedatabaseonthescaleofMegaFacecouldneverbeestablishedagain.[192]InSeptember2019theSwedishDataProtectionAuthority(DPA)issueditsfirsteverfinancialpenaltyforaviolationoftheEU'sGeneralDataProtectionRegulation(GDPR)againstaschoolthatwasusingthetechnologytoreplacetime-consumingrollcallsduringclass.TheDPAfoundthattheschoolillegallyobtainedthebiometricdataofitsstudentswithoutcompletinganimpactassessment.InadditiontheschooldidnotmaketheDPAawareofthepilotscheme.A200,000SEKfine(€19,000/$21,000)wasissued.[193] IntheUnitedStatesofAmericaseveralU.S.stateshavepassedlawstoprotecttheprivacyofbiometricdata.ExamplesincludetheIllinoisBiometricInformationPrivacyAct(BIPA)andtheCaliforniaConsumerPrivacyAct(CCPA).[194]InMarch2020CaliforniaresidentsfiledaclassactionagainstClearviewAI,allegingthatthecompanyhadillegallycollectedbiometricdataonlineandwiththehelpoffacerecognitiontechnologybuiltupadatabaseofbiometricdatawhichwassoldtocompaniesandpoliceforces.AtthetimeClearviewAIalreadyfacedtwolawsuitsunderBIPA[195]andaninvestigationbythePrivacyCommissionerofCanadaforcompliancewiththePersonalInformationProtectionandElectronicDocumentsAct(PIPEDA).[196] Bansontheuseoffacialrecognitiontechnology[edit] InMay2019,SanFrancisco,CaliforniabecamethefirstmajorUnitedStatescitytobantheuseoffacialrecognitionsoftwareforpoliceandotherlocalgovernmentagencies'usage.[197]SanFranciscoSupervisor,AaronPeskin,introducedregulationsthatwillrequireagenciestogainapprovalfromtheSanFranciscoBoardofSupervisorstopurchasesurveillancetechnology.[198]Theregulationsalsorequirethatagenciespubliclydisclosetheintendedusefornewsurveillancetechnology.[198]InJune2019,Somerville,MassachusettsbecamethefirstcityontheEastCoasttobanfacesurveillancesoftwareforgovernmentuse,[199]specificallyinpoliceinvestigationsandmunicipalsurveillance.[200]InJuly2019,Oakland,Californiabannedtheusageoffacialrecognitiontechnologybycitydepartments.[201] TheAmericanCivilLibertiesUnion("ACLU")hascampaignedacrosstheUnitedStatesfortransparencyinsurveillancetechnology[200]andhassupportedbothSanFranciscoandSomerville'sbanonfacialrecognitionsoftware.TheACLUworkstochallengethesecrecyandsurveillancewiththistechnology.[citationneeded][202] InJanuary2020,theEuropeanUnionsuggested,butthenquicklyscrapped,aproposedmoratoriumonfacialrecognitioninpublicspaces.[203][204] DuringtheGeorgeFloydprotests,useoffacialrecognitionbycitygovernmentwasbannedinBoston,Massachusetts.[205]AsofJune10,2020,municipalusehasbeenbannedin:[6] Berkeley,California Oakland,California Boston,Massachusetts–June30,2020[206] Brookline,Massachusetts Cambridge,Massachusetts Northampton,Massachusetts Springfield,Massachusetts Somerville,Massachusetts Portland,Oregon–September2020[207] TheWestLafayette,IndianaCityCouncilpassedanordinancebanningfacialrecognitionsurveillancetechnology.[208] OnOctober27,2020,22humanrightsgroupscalledupontheUniversityofMiamitobanfacialrecognitiontechnology.Thiscameafterthestudentsaccusedtheschoolofusingthesoftwaretoidentifystudentprotesters.Theallegationswere,however,deniedbytheuniversity.[209] TheEuropean"ReclaimYourFace"coalitionlaunchedinOctober2020.ThecoalitioncallsforabanonfacialrecognitionandlaunchedaEuropeanCitizens'InitiativeinFebruary2021.Morethan60organizationscallontheEuropeanCommissiontostrictlyregulatetheuseofbiometricsurveillancetechnologies.[210] AstatepolicereformlawinMassachusettswilltakeeffectinJuly2021;abanpassedbythelegislaturewasrejectedbygovernorCharlieBaker.[211]Instead,thelawrequiresajudicialwarrant,limitthepersonnelwhocanperformthesearch,recorddataabouthowthetechnologyisused,andcreateacommissiontomakerecommendationsaboutfutureregulations.[212] Emotionrecognition[edit] Inthe18thand19thcentury,thebeliefthatfacialexpressionsrevealedthemoralworthortrueinnerstateofahumanwaswidespreadandphysiognomywasarespectedscienceintheWesternworld.Fromtheearly19thcenturyonwardsphotographywasusedinthephysiognomicanalysisoffacialfeaturesandfacialexpressiontodetectinsanityanddementia.[213]Inthe1960sand1970sthestudyofhumanemotionsanditsexpressionswasreinventedbypsychologists,whotriedtodefineanormalrangeofemotionalresponsestoevents.[214]Theresearchonautomatedemotionrecognitionhassincethe1970sfocusedonfacialexpressionsandspeech,whichareregardedasthetwomostimportantwaysinwhichhumanscommunicateemotionstootherhumans.Inthe1970stheFacialActionCodingSystem(FACS)categorizationforthephysicalexpressionofemotionswasestablished.[215]ItsdeveloperPaulEkmanmaintainsthattherearesixemotionsthatareuniversaltoallhumanbeingsandthatthesecanbecodedinfacialexpressions.[216]Researchintoautomaticemotionspecificexpressionrecognitionhasinthepastdecadesfocusedonfrontalviewimagesofhumanfaces.[217] In2016,facialfeatureemotionrecognitionalgorithmswereamongthenewtechnologies,alongsidehigh-definitionCCTV,highresolution3Dfacerecognitionandirisrecognition,thatfoundtheirwayoutofuniversityresearchlabs.[218]In2016,FacebookacquiredFacioMetrics,afacialfeatureemotionrecognitioncorporatespin-offbyCarnegieMellonUniversity.InthesameyearAppleInc.acquiredthefacialfeatureemotionrecognitionstart-upEmotient.[219]Bytheendof2016,commercialvendorsoffacialrecognitionsystemsofferedtointegrateanddeployemotionrecognitionalgorithmsforfacialfeatures.[220]TheMIT'sMediaLabspin-offAffectiva[221]bylate2019offeredafacialexpressionemotiondetectionproductthatcanrecognizeemotionsinhumanswhiledriving.[219] Anti-facialrecognitionsystems[edit] Thedevelopmentofanti-facialrecognitiontechnologyiseffectivelyanarmsracebetweenprivacyresearchersandbigdatacompanies.BigdatacompaniesincreasinglyuseconvolutionalAItechnologytocreateevermoreadvancedfacialrecognitionmodels.Solutionstoblockfacialrecognitionmaynotworkonnewersoftware,orondifferenttypesoffacialrecognitionmodels.Onepopularcitedexampleoffacial-recognitionblockingistheCVDazzlemakeupandhaircutsystem,butthecreatorsnoteontheirwebsitethatithasbeenoutdatedforquitesometimeasitwasdesignedtocombataparticularfacialrecognitionalgorithmandmaynotwork.[222]Anotherexampleistheemergenceoffacialrecognitionthatcanidentifypeoplewearingfacemasksandsunglasses,especiallyaftertheCOVID-19pandemic.[223] Giventhatbigdatacompanieshavemuchmorefundingthanprivacyresearchers,itisverydifficultforanti-facialrecognitionsystemstokeepup.Thereisalsonoguaranteethatobfuscationtechniquesthatwereusedforimagestakeninthepastandstored,suchasmasksorsoftwareobfuscation,wouldprotectusersfromfacial-recognitionanalysisofthoseimagesbyfuturetechnology.[224] InJanuary2013,JapaneseresearchersfromtheNationalInstituteofInformaticscreated'privacyvisor'glassesthatusenearlyinfraredlighttomakethefaceunderneathitunrecognizabletofacerecognitionsoftwarethatuseinfrared.[225]Thelatestversionusesatitaniumframe,light-reflectivematerialandamaskwhichusesanglesandpatternstodisruptfacialrecognitiontechnologythroughbothabsorbingandbouncingbacklightsources.[226][227][228][229]However,thesemethodsareusedtopreventinfraredfacialrecognitionandwouldnotworkonAIfacialrecognitionofplainimages.Someprojectsuseadversarialmachinelearningtocomeupwithnewprintedpatternsthatconfuseexistingfacerecognitionsoftware.[230] Onemethodthatmayworktoprotectfromfacialrecognitionsystemsarespecifichaircutsandmake-uppatternsthatpreventtheusedalgorithmstodetectaface,knownascomputervisiondazzle.[222]Incidentally,themakeupstylespopularwithJuggalosmayalsoprotectagainstfacialrecognition.[231] Facialmasksthatareworntoprotectfromcontagiousvirusescanreducetheaccuracyoffacialrecognitionsystems.A2020NISTstudy,testedpopularone-to-onematchingsystemsandfoundafailureratebetweenfiveandfiftypercentonmaskedindividuals.TheVergespeculatedthattheaccuracyrateofmasssurveillancesystems,whichwerenotincludedinthestudy,wouldbeevenlessaccuratethantheaccuracyofone-to-onematchingsystems.[232]ThefacialrecognitionofApplePaycanworkthroughmanybarriers,includingheavymakeup,thickbeardsandevensunglasses,butfailswithmasks.[233]However,facialrecognitionofmaskedfacesisincreasinglygettingmorereliable. Anothersolutionistheapplicationofobfuscationtoimagesthatmayfoolfacialrecognitionsystemswhilestillappearingnormaltoahumanuser.Thesecouldbeusedforwhenimagesarepostedonlineoronsocialmedia.However,asitishardtoremoveimagesoncetheyareontheinternet,theobfuscationontheseimagesmaybedefeatedandthefaceoftheuseridentifiedbyfutureadvancesintechnology.Twoexamplesofthistechnique,developedin2020,aretheANU's'CameraAdversaria'cameraapp,andtheUniversityofChicago'sFawkesalgorithmwhichappliesobfuscationtoalreadytakenphotos.[224]However,by2021theFawkesobfuscationalgorithmhadalreadybeenspecificallytargetedbyMicrosoftAzurewhichchangedits'algorithmtolowerFawkes'effectiveness.[234] Seealso[edit] AIeffect AmazonRekognition Applicationsofartificialintelligence Artificialintelligenceforvideosurveillance Automaticnumberplaterecognition Biometrictechnologyinaccesscontrol CokeZeroFacialProfiler Computerprocessingofbodylanguage Computervision DeepFace Faceperception FaceRecognitionGrandChallenge FindFace GlasgowFaceMatchingTest ISO/IEC19794-5 MALINTENT Nationalbiometricidcard Multimediainformationretrieval Multilinearsubspacelearning Patternrecognition,analogyandcase-basedreasoning Retinalscan SenseTime Superrecognisers Templatematching Three-dimensionalfacerecognition Veinmatching Gaitanalysis Lists Listofcomputervisiontopics ListofemergingtechnologiesOutlineofartificialintelligence References[edit] ^Thorat,S.B.;Nayak,S.K.;JyotiPDandale(2010)."FacialRecognitionTechnology:AnanalysiswithscopeinIndia".arXiv:1005.4263[cs.MA]. ^Chen,S.K;Chang,Y.H(2014).2014InternationalConferenceonArtificialIntelligenceandSoftwareEngineering(AISE2014).DEStechPublications,Inc.p. 21.ISBN 9781605951508. ^Bramer,Max(2006).ArtificialIntelligenceinTheoryandPractice:IFIP19thWorldComputerCongress,TC12:IFIPAI2006Stream,August21–24,2006,Santiago,Chile.Berlin:SpringerScience+BusinessMedia.p. 395.ISBN 9780387346540. ^"FacialRecognitionTechnology:FederalLawEnforcementAgenciesShouldHaveBetterAwarenessofSystemsUsedByEmployees".www.gao.gov.RetrievedSeptember5,2021. ^Wiggers,Kyle(March5,2022)."Studywarnsdeepfakescanfoolfacialrecognition".VentureBeat.RetrievedJune4,2022. ^ab"IBMbowsoutoffacialrecognitionmarket-".GCN.June10,2020.RetrievedOctober7,2021. ^RachelMetz."Facebookisshuttingdownitsfacialrecognitionsoftware".CNN.RetrievedNovember5,2021. ^Hill,Kashmir;Mac,Ryan(November2,2021)."Facebook,CitingSocietalConcerns,PlanstoShutDownFacialRecognitionSystem".TheNewYorkTimes.ISSN 0362-4331.RetrievedNovember5,2021. ^Nilsson,NilsJ.(2009).TheQuestforArtificialIntelligence.CambridgeUniversityPress.ISBN 9781139642828. ^deLeeuw,Karl;Bergstra,Jan(2007).TheHistoryofInformationSecurity:AComprehensiveHandbook.Elsevier.p. 266.ISBN 9780444516084. ^Gates,Kelly(2011).OurBiometricFuture:FacialRecognitionTechnologyandtheCultureofSurveillance.NYUPress.pp. 48–49.ISBN 9780814732090. ^Gates,Kelly(2011).OurBiometricFuture:FacialRecognitionTechnologyandtheCultureofSurveillance.NYUPress.pp. 49–50.ISBN 9780814732090. ^Gates,Kelly(2011).OurBiometricFuture:FacialRecognitionTechnologyandtheCultureofSurveillance.NYUPress.p. 52.ISBN 9780814732090. ^Gates,Kelly(2011).OurBiometricFuture:FacialRecognitionTechnologyandtheCultureofSurveillance.NYUPress.p. 53.ISBN 9780814732090. ^Gates,Kelly(2011).OurBiometricFuture:FacialRecognitionTechnologyandtheCultureofSurveillance.NYUPress.p. 54.ISBN 9780814732090. ^MalayK.Kundu;SushmitaMitra;DebasisMazumdar;SankarK.Pal,eds.(2012).PerceptionandMachineIntelligence:FirstIndo-JapanConference,PerMIn2012,Kolkata,India,January12–13,2011,Proceedings.SpringerScience&BusinessMedia.p. 29.ISBN 9783642273865. ^Wechsler,Harry(2009).MalayK.Kundu;SushmitaMitra(eds.).ReliableFaceRecognitionMethods:SystemDesign,ImplementationandEvaluation.SpringerScience&BusinessMedia.pp. 11–12.ISBN 9780387384641. ^JunWang;LaiwanChan;DeLiangWang,eds.(2012).NeuralInformationProcessing:13thInternationalConference,ICONIP2006,HongKong,China,October3–6,2006,Proceedings,PartII.SpringerScience&BusinessMedia.p. 198.ISBN 9783540464822. ^Wechsler,Harry(2009).ReliableFaceRecognitionMethods:SystemDesign,ImplementationandEvaluation.SpringerScience&BusinessMedia.p. 12.ISBN 9780387384641. ^Wechsler,Harry(2009).MalayK.Kundu;SushmitaMitra(eds.).ReliableFaceRecognitionMethods:SystemDesign,ImplementationandEvaluation.SpringerScience&BusinessMedia.p. 12.ISBN 9780387384641. ^MalayK.Kundu;SushmitaMitra;DebasisMazumdar;SankarK.Pal,eds.(2012).PerceptionandMachineIntelligence:FirstIndo-JapanConference,PerMIn2012,Kolkata,India,January12–13,2011,Proceedings.SpringerScience&BusinessMedia.p. 29.ISBN 9783642273865. ^"MugspotCanFindAFaceInTheCrowd–Face-RecognitionSoftwarePreparesToGoToWorkInTheStreets".ScienceDaily.November12,1997.RetrievedNovember6,2007. ^MalayK.Kundu;SushmitaMitra;DebasisMazumdar;SankarK.Pal,eds.(2012).PerceptionandMachineIntelligence:FirstIndo-JapanConference,PerMIn2012,Kolkata,India,January12–13,2011,Proceedings.SpringerScience&BusinessMedia.p. 29.ISBN 9783642273865. ^Li,StanZ.;Jain,AnilK.(2005).HandbookofFaceRecognition.SpringerScience&BusinessMedia.pp. 14–15.ISBN 9780387405957. ^KumarDatta,Asit;Datta,Madhura;KumarBanerjee,Pradipta(2015).FaceDetectionandRecognition:TheoryandPractice.CRC.p. 123.ISBN 9781482226577. ^Severi,Misty(April15,2022)."UkraineusesfacialrecognitionsoftwaretoidentifydeadRussiansoldiers". ^"Facialrecognitiontechnologyisavaluabletool".LosAngelesDailyNews.May15,2022. ^Italiano,Laura(April15,2022)."UkraineisusingfacialrecognitiontoIDdeadRussiansoldiersandsendphotosofcorpseshometotheirmoms:report".BusinessInsider. ^Li,StanZ.;Jain,AnilK.(2005).HandbookofFaceRecognition.SpringerScience&BusinessMedia.p. 1.ISBN 9780387405957. ^Li,StanZ.;Jain,AnilK.(2005).HandbookofFaceRecognition.SpringerScience&BusinessMedia.p. 2.ISBN 9780387405957. ^"AirportFacialRecognitionPassengerFlowManagement".hrsid.com. ^abcBonsor,K.(September4,2001)."HowFacialRecognitionSystemsWork".RetrievedJune2,2008. ^Smith,Kelly."FaceRecognition"(PDF).RetrievedJune4,2008. ^R.BrunelliandT.Poggio,"FaceRecognition:FeaturesversusTemplates",IEEETrans.onPAMI,1993,(15)10:1042–1052 ^R.Brunelli,TemplateMatchingTechniquesinComputerVision:TheoryandPractice,Wiley,ISBN 978-0-470-51706-2,2009([1]TMbook) ^Zhang,David;Jain,Anil(2006).AdvancesinBiometrics:InternationalConference,ICB2006,HongKong,China,January5–7,2006,Proceedings.Berlin:SpringerScience&BusinessMedia.p. 183.ISBN 9783540311119. ^"AStudyontheDesignandImplementationofFacialRecognitionApplicationSystem".InternationalJournalofBio-ScienceandBio-Technology. ^H.Ugail,Deepfacerecognitionusingfullandpartialfaceimages,Elesevier,ISBN 978-0-12-822109-9,2022(https://doi.org/10.1016/B978-0-12-822109-9.00015-1]AdvancedMethodsandDeepLearninginComputerVision) ^HarryWechsler(2009).ReliableFaceRecognitionMethods:SystemDesign,ImplementationandEvaluation.SpringerScience&BusinessMedia.p. 196.ISBN 9780387384641. ^abcdWilliams,Mark."BetterFace-RecognitionSoftware".RetrievedJune2,2008. ^Crawford,Mark."Facialrecognitionprogressreport".SPIENewsroom.RetrievedOctober6,2011. ^Kimmel,Ron."Three-dimensionalfacerecognition"(PDF).RetrievedJanuary1,2005. ^Duhn,S.von;Ko,M.J.;Yin,L.;Hung,T.;Wei,X.(September1,2007)."Three-ViewSurveillanceVideoBasedFaceModelingforRecogniton[sic]".Three-ViewSurveillanceVideoBasedFaceModelingforRecognition.pp. 1–6.doi:10.1109/BCC.2007.4430529.ISBN 978-1-4244-1548-9.S2CID 25633949. ^abSocolinsky,DiegoA.;Selinger,Andrea(January1,2004)."ThermalFaceRecognitioninanOperationalScenario".CVPR'04.IEEEComputerSociety.pp. 1012–1019–viaACMDigitalLibrary. ^ab"ArmyBuildsFaceRecognitionTechnologythatWorksinLow-LightConditions".AZoRobotics.April18,2018.RetrievedAugust17,2018. ^ThirimachosBourlai(2016).FaceRecognitionAcrosstheImagingSpectrum.Springer.p. 142.ISBN 9783319285016. ^ThirimachosBourlai(2016).FaceRecognitionAcrosstheImagingSpectrum.Springer.p. 140.ISBN 9783319285016. ^"Armydevelopsfacerecognitiontechnologythatworksinthedark".ArmyResearchLaboratory.April16,2018.RetrievedAugust17,2018. ^abRiggan,Benjamin;Short,Nathaniel;Hu,Shuowen(March2018)."ThermaltoVisibleSynthesisofFaceImagesusingMultipleRegions".ResearchGate.arXiv:1803.07599.Bibcode:2018arXiv180307599R. ^Cole,Sally(June2018)."U.S.Army'sAIfacialrecognitionworksinthedark".MilitaryEmbeddedSystems.p. 8. ^Shontell,Alyson(September15,2015)."SnapchatbuysLooksery,a2-year-oldstartupthatletsyouPhotoshopyourfacewhileyouvideochat".BusinessInsiderSingapore.RetrievedApril9,2018. ^KumarMandal,Jyotsna;Bhattacharya,Debika(2019).EmergingTechnologyinModellingandGraphics:ProceedingsofIEMGraph2018.Springer.p. 672.ISBN 9789811374036. ^Simonite,Tom."FacebookCreatesSoftwareThatMatchesFacesAlmostasWellasYouDo".MITTechnologyReview.RetrievedApril9,2018. ^"Facebook'sDeepFaceshowsseriousfacialrecognitionskills".RetrievedApril9,2018. ^"WhyFacebookisbeatingtheFBIatfacialrecognition".TheVerge.RetrievedApril9,2018. ^"HowTikTok's'ForYou'AlgorithmActuallyWorks".Wired.ISSN 1059-1028.RetrievedApril17,2021. ^"HowTikTokrecommendsvideos#ForYou".TikTok.June18,2020.ArchivedfromtheoriginalonJune18,2020.RetrievedApril22,2021. ^"TikTokagreeslegalpayoutoverfacialrecognition".BBCNews.February26,2021.ArchivedfromtheoriginalonFebruary26,2021.RetrievedApril22,2021. ^"Aglimpseatbankbranchesofthefuture:videowalls,booth-sizedlocationsand24/7access".USAToday.RetrievedAugust13,2018. ^Heater,Brian."Don'trelyonFaceUnlocktokeepyourphone secure".TechCrunch.RetrievedNovember2,2017. ^"GalaxyS8facerecognitionalreadydefeatedwithasimplepicture".ArsTechnica.RetrievedNovember2,2017. ^"HowFacialRecognitionWorksinXboxKinect".Wired.RetrievedNovember2,2017. ^"Windows10says"Hello"tologginginwithyourfaceandtheendofpasswords".ArsTechnica.March17,2015.RetrievedMarch17,2015. ^Kubota,Yoko(September27,2017)."AppleiPhoneXProductionWoeSparkedbyJulietandHerRomeo".TheWallStreetJournal.ArchivedfromtheoriginalonSeptember28,2017.RetrievedSeptember27,2017. ^Kubota,Yoko(September27,2017)."AppleiPhoneXProductionWoeSparkedbyJulietandHerRomeo".TheWallStreetJournal.ISSN 0099-9660.RetrievedApril10,2018. ^ab"ThefivebiggestquestionsaboutApple'snewfacialrecognitionsystem".TheVerge.RetrievedApril10,2018. ^"Apple'sFaceIDFeatureWorksWithMostSunglasses,CanBeQuicklyDisabledtoThwartThieves".RetrievedApril10,2018. ^Heisler,Yoni(November3,2017)."InfraredvideoshowsofftheiPhoneX'snewFaceIDfeatureinaction".BGR.RetrievedApril10,2018. ^"Modigovtnowplansa'touchless'vaccinationprocess,withAadhaar-basedfacialrecognition".ThePrint.April6,2021.RetrievedFebruary12,2022. ^"DespitePrivacyFears,Aadhaar-LinkedFacialRecognitionUsedForCovid-19Vaccines".Inc42.April7,2021.RetrievedFebruary12,2022. ^"JointStatement:SaynotoAadhaarbasedFacialRecognitionforVaccination!".InternetFreedomFoundation.April14,2021.RetrievedFebruary12,2022. ^"PanopticTracker,Finance(PensionCell)Department,GovernmentofMeghalaya".PanopticProject.RetrievedFebruary12,2022. ^"Meghalayaclarifiesoncontroversialapp:'FacialRecognitionTechnologydoesn'trequireanyanchoringlegislation'".IndianExpress.November18,2021.RetrievedFebruary12,2022. ^"SmartgatesFaceeditingfortheminsofthecanwehave".AustralianBorderForce.RetrievedMarch11,2019. ^"Ourhistory".NewZealandCustomsService.RetrievedMarch11,2019. ^"FacialrecognitiontechnologyiscomingtoCanadianairportsthisspring".CBCNews.RetrievedMarch3,2017. ^abc"FaceOff:ThelawlessgrowthoffacialrecognitioninUKpolicing"(PDF).BigBrotherWatch. ^Anthony,Sebastian(June6,2017)."UKpolicearrestmanviaautomaticface-recognitiontech".ArsTechnica. ^abRees,Jenny(September4,2019)."Policeuseoffacialrecognitionruledlawful".RetrievedNovember8,2019. ^Burgess,Matt(January24,2020)."TheMetPolicewillstartusinglivefacialrecognitionacrossLondon".WiredUK.ISSN 1357-0978.RetrievedJanuary24,2020. ^DanicaKirka(August11,2020)."UKcourtsaysfacerecognitionviolateshumanrights".TechPlore.RetrievedOctober4,2020. ^Fortune."Here'sHowManyAdultFacesAreScannedFromFacialRecognitionDatabases". ^ab"Thetroublewithfacialrecognitiontechnology(intherealworld)". ^"Real-TimeFacialRecognitionIsAvailable,ButWillU.S.PoliceBuyIt?".NPR.org.NPR. ^"PoliceFacialRecognitionDatabasesLogAboutHalfOfAmericans".NPR.org.NPR. ^Knezevich,KevinRector,Alison."Maryland'suseoffacialrecognitionsoftwarequestionedbyresearchers,civillibertiesadvocates". ^"NextGenerationIdentification".FBI.RetrievedApril5,2016. ^abICEUsesFacialRecognitionToSiftStateDriver'sLicenseRecords,ResearchersSay ^"TSAhadexpresseditsintentiontoadoptasimilarprogramfordomesticairtravel".USAToday.August16,2019. ^Shen,Xinmei(October4,2018).""Skynet",China'smassivevideosurveillancenetwork".SouthChinaMorningPost.RetrievedDecember13,2020. ^Chan,TaraFrancis(March27,2018)."16partsofChinaarenowusingSkynet".BusinessInsider.RetrievedDecember13,2020. ^"Fromaletojail:facialrecognitioncatchescriminalsatChinabeerfestival".TheGuardian.September1,2017.RetrievedMarch8,2018. ^"PoliceusefacialrecognitiontechnologytodetectwantedcriminalsduringbeerfestivalinChinesecityofQingdao".opengovasia.com.OpenGovAsia.ArchivedfromtheoriginalonNovember16,2017.RetrievedMarch8,2018. ^"Chinesepoliceareusingsmartglassestoidentifypotentialsuspects".TechCrunch.February8,2018.RetrievedDecember3,2020. ^"Beijingpoliceareusingfacial-recognitionglassestoidentifycarpassengersandnumberplates".BusinessInsider.March12,2018.RetrievedDecember3,2020. ^"China'smassiveinvestmentinartificialintelligencehasaninsidiousdownside".ScienceAAAS.February7,2018.RetrievedFebruary23,2018. ^"Chinabetsonfacialrecognitioninbigdrivefortotalsurveillance".TheWashingtonPost.2018.RetrievedFebruary23,2018. ^Liao,Rita(May8,2019)."Alibaba-backedfacialrecognitionstartupMegviiraises$750million".TechCrunch.RetrievedAugust28,2019. ^Dai,Sarah(June5,2019)."AIunicornMegviinotbehindappusedforsurveillanceinXinjiang,sayshumanrightsgroup".SouthChinaMorningPost.RetrievedAugust28,2019. ^ChengLeng,YingzhiYangandRyanWoo(February20,2020)."Exclusive:HundredsofChinesebusinessesseekbillionsinloanstocontendwithcoronavirus".Reuters.RetrievedOctober5,2020.{{citenews}}:CS1maint:usesauthorsparameter(link) ^"Alawsuitagainstface-scansinChinacouldhavebigconsequences".TheEconomist.November9,2019. ^Xiaoshan,Huang;Wen,Cheng."NewevidenceshowingTencentmonitorsoverseasusers".ArchivedfromtheoriginalonAugust16,2020.RetrievedAugust15,2020. ^ZakDoffman(August26,2019)."HongKongExposesBothSidesOfChina'sRelentlessFacialRecognitionMachine".Forbes.RetrievedDecember3,2020.{{citenews}}:CS1maint:usesauthorsparameter(link) ^Lohr,Steve(February9,2018)."FacialRecognitionIsAccurate,ifYou'reaWhiteGuy".TheNewYorkTimes.ISSN 0362-4331.RetrievedFebruary14,2022. ^"NCRB'sNationalAutomatedFacialRecognitionSystem".panoptic.in.RetrievedFebruary14,2022. ^ab"WatchtheWatchmenSeriesPart4 :TheNationalAutomatedFacialRecognitionSystem".InternetFreedomFoundation.October7,2020.RetrievedFebruary14,2022. ^"JusticeK.S.Puttaswamy(Retd)vsUnionOfIndiaon26September,2018".IndianKanoon. ^"Wemightbeinthemarketforanewkindoffacemask".InternetFreedomFoundation.July18,2019.RetrievedFebruary14,2022. ^Barik,Soumyarendra(October22,2019)."'Fingerprintisnotabigissue':Hyderabadpoliceoncollectingbiometricsof'suspects'".MediaNama.RetrievedFebruary14,2022. ^ab"AutomatedFacialRecognitionTechnology(MoratoriumandReview)Bill[HL]-ParliamentaryBills-UKParliament".bills.parliament.uk.RetrievedSeptember10,2021. ^"UPPolicelaunch'Trinetra',itsAI-poweredfacerecognitionapptocatchcriminals".TheFinancialExpress.RetrievedFebruary14,2022. ^"UttarakhandPoliceacquirefacerecognitionsoftwaretohelpnabcriminals".HindustanTimes.August27,2018.RetrievedFebruary14,2022. ^"CrimeandCriminalTrackingNetwork&Systems(CCTNS)|NationalCrimeRecordsBureau".ncrb.gov.in.RetrievedFebruary18,2022. ^"CCTNSProjecttoletpolicestations'talk':whereitstands,andhowitcanhelpfightcrime".TheIndianExpress.November20,2015.RetrievedFebruary18,2022. ^"HomeMinistrymovestogetautomatedfacialrecognitionsystemforpolice".HindustanTimes.July8,2019.RetrievedFebruary18,2022. ^"IFF'sLegalNoticetotheNCRBontheRevisedRFPfortheNationalAutomatedFacialRecognitionSystem".InternetFreedomFoundation.July15,2020.RetrievedFebruary18,2022. ^abcdefgPARLIAMENTOFINDIA.RAJYASABHA.TWOHUNDREDTHIRTYSEVENTHREPORTONPOLICE-TRAINING,MODERNISATIONANDREFORMS(PDF).DEPARTMENT-RELATEDPARLIAMENTARYSTANDINGCOMMITTEEONHOMEAFFAIRS,India.2022.p. 34. ^U.SudhakarReddy(November10,2021)."8.3lakhcamerasinTelangana,Hyderabadturningintosurveillancecity:Amnesty|HyderabadNews-TimesofIndia".TheTimesofIndia.RetrievedFebruary18,2022. ^"Indiangovt'sapproachtofacialrecognitionisflawed&drivenbyfaultyassumptions".ThePrint.November27,2019.RetrievedFebruary15,2022. ^"RighttoInformationUpdatesfromDelhiPolice,KolkataPoliceandTelanganaStateTechnologyServices".panoptic.in.RetrievedFebruary15,2022. ^"Section8(1)(d)inTheRightToInformationAct,2005".IndianKanoon. ^"ProjectPanoptic:RTIUpdatesfromDelhiPolice,KolkataPoliceandTelanganaStateTechnologyServices".InternetFreedomFoundation.December1,2020.RetrievedFebruary15,2022. ^Chunduru,Aditya(December2,2020)."RTI:Kolkata,Delhipolicerefusetogiveinformationonfacialrecognitionsystems".MediaNama.RetrievedFebruary15,2022. ^"MexicanGovernmentAdoptsFaceItFaceRecognitionTechnologytoEliminateDuplicateVoterRegistrationsinUpcomingPresidentialElection".BusinessWire.May11,2000.ArchivedfromtheoriginalonMarch5,2016.RetrievedJune2,2008. ^abcSelinger,Evan;Polonetsky,Jules;Tene,Omer(2018).TheCambridgeHandbookofConsumerPrivacy.CambridgeUniversityPress.p. 112.ISBN 9781316859278. ^Vogel,Ben."Panamaputsnamestomorefaces".IHSJane'sAirportReview.ArchivedfromtheoriginalonOctober12,2014.RetrievedOctober7,2014. ^"'Made-in-China'productsshineatRioOlympics".TheStateCouncil,Thepeople'sRepublicofChina.August15,2016.RetrievedNovember14,2020. ^Kayser-Bril,Nicolas(December11,2019)."Atleast11policeforcesusefacerecognitionintheEU,AlgorithmWatchreveals".AlgorithmWatch. ^Pedriti,Corina(January28,2021)."FlushwithEUfunds,Greekpolicetointroducelivefacerecognitionbeforethesummer".AlgorithmWatch. ^Coluccini,Riccardo(January13,2021)."LoscontroViminale-Garantedellaprivacysulriconoscimentofaccialeintemporeale".IrpiMedia. ^Techredacteur,JoostSchellevis."Politiegaatverdachtenopsporenmetgezichtsherkenning".nos.nl(inDutch).RetrievedSeptember22,2019. ^Boon,Lex(August25,2018)."Meekijkenmetde226gemeentecamera's".HetParool(inDutch).RetrievedSeptember22,2019. ^HowCCTVsurveillanceposesathreattoprivacyinSouthAfrica ^Ross,Tim(2007)."3VRFeaturedonFoxBusinessNews".MoneyforBreakfast(Interview).FoxBusiness.Interviewer:Now,canIbuysomethinglikethis?Isthis...doyoureallyrestrictthecustomersforthis?TimRoss:It'sprimarilybeingpurchasedbybanks,retailers,andthegovernmenttodayandissoldthroughavarietyofsecuritychannels. ^"ImproveCustomerService".3VR.ArchivedfromtheoriginalonAugust14,2012.3VR'sVideoIntelligencePlatform(VIP)transformscustomerservicebyallowingbusinessesto:•Optimizestaffingdecisions,increasesalesconversionratesanddecreasecustomerwaittimesbybringingextraordinaryclaritytotheanalysisoftrafficpatterns•Alignstaffingdecisionswithactualcustomeractivity,usingdwellandqueuelineanalyticstodecreasecustomerwaittimes•Increasecompetitivenessbyusing3VR'sfacialsurveillanceanalytictofacilitatepersonalizedcustomergreetingsbyemployees•Createloyaltyprogramsbycombiningpointofsale(POS)datawithfacialrecognition ^abcdefghDastin,JeffreyL.;Cadell,Cate;Yang,Yizhing;Tham,Engen;Goh,Brenda;Master,Farah;Jackson,Lucas;Michalska,Aleksandra;Hart,Samuel(July28,2020).Marquis,Julie;Robinson,Simon(eds.)."SpecialReport:RiteAiddeployedfacialrecognitionsystemsinhundredsofU.S.stores".U.S.LegalNews.Reuters.FurtherreportingbyPareshDave,TomBergin,andtheReutersBeijingandShanghainewsrooms;dataanalysisbyRyanMcNeill.ArchivedfromtheoriginalonDecember19,2020. ^abMayhew,Stephen(March17,2019)."Casinosdownunderdeployfacialrecognitiontechtospotoffenders,problemgamblers|BiometricUpdate".www.biometricupdate.com.RetrievedJune30,2022. ^Scanlan,Rebekah(June29,2022)."TheGoodGuysscrap'creepy'camerafeatureafterbacklash".News.com.au. ^Greene,Lisa(February15,2001)."Facescansmatchfewsuspects"(SHTML).St.PetersburgTimes.ArchivedfromtheoriginalonNovember30,2014.RetrievedJune30,2011.ByusingViisagesoftware,policematched19people'sfacestophotosofpeoplearrestedinthepastforminorpickpocketing,fraudandothercharges.Theyweren'tchargedwithanygame-daymisdeeds.THISISAFARCE ^abcKrause,Mike(January14,2002)."Isfacerecognitionjusthigh-techsnakeoil?".EnterStageRight.ISSN 1488-1756.ArchivedfromtheoriginalonJanuary24,2002.RetrievedJune30,2011. ^"Windows10'sPhotosappisgettingsmarterimagesearchjustlikeGooglePhotos".TheVerge.RetrievedNovember2,2017. ^Perez,Sarah."GooglePhotosupgradedwithnewsharingfeatures,photobooks,andGoogle Lens".TechCrunch.RetrievedNovember2,2017. ^"FaceRecognitionApplications".Animetrics.ArchivedfromtheoriginalonJuly13,2008.RetrievedJune4,2008. ^Giaritelli,Anna(December13,2018)."TaylorSwiftusedairport-stylefacialrecognitiononconcertgoers".washingtonexaminer.com.RetrievedDecember13,2018. ^"ManchesterCitytriesfacialrecognitiontobeatfootballqueues".TheTimes.RetrievedAugust18,2019. ^"ManchesterCitywarnedagainstusingfacialrecognitiononfans".TheGuardian.RetrievedAugust18,2019. ^Olson,Parmy(August1,2020)."FacialRecognition'sNextBigPlay:theSportsStadium".TheWallStreetJournal.ISSN 0099-9660.RetrievedAugust3,2020. ^"FacialRecognitionTechnologyTest".WaltDisneyWorldParkEntryTechnologyTest.Disney.ArchivedfromtheoriginalonApril22,2021.RetrievedApril22,2021. ^R.KimmelandG.Sapiro(April30,2003)."TheMathematicsofFaceRecognition".SIAMNews.ArchivedfromtheoriginalonJuly15,2007.RetrievedApril30,2003. ^abc"TopFiveBiometrics:Face,Fingerprint,Iris,PalmandVoice".Bayometric.January23,2017.RetrievedApril10,2018. ^"PrivacyPrinciplesforFacialRecoginitionTechnologyinCommercialApplications"(PDF).fpf.org. ^Haghighat,Mohammad;Abdel-Mottaleb,Mohamed(2017)."LowResolutionFaceRecognitioninSurveillanceSystemsUsingDiscriminantCorrelationAnalysis".201712thIEEEInternationalConferenceonAutomaticFace&GestureRecognition(FG2017).pp. 912–917.doi:10.1109/FG.2017.130.ISBN 978-1-5090-4023-0.S2CID 36639614. ^"PassportCanada–Photos".passportcanada.gc.ca.ArchivedfromtheoriginalonMarch1,2009. ^Albiol,A.,Albiol,A.,Oliver,J.,Mossi,J.M.(2012).Whoiswhoatdifferentcameras:peoplere-identificationusingdepthcameras.ComputerVision,IET.Vol6(5),378–387. ^abcd"Facialrecognitionsoftwarehasagenderproblem".nsf.gov.RetrievedMay9,2021. ^Rehnman,Jenny."Theroleofgenderinfacerecognition"(PDF).www.diva-portal.org.p. 69. ^Mishra,MarutiV.;Likitlersuang,Jirapat;Wilmer,JeremyB.;Cohan,Sarah;Germine,Laura;DeGutis,JosephM.(November29,2019)."GenderDifferencesinFamiliarFaceRecognitionandtheInfluenceofSocioculturalGenderInequality".ScientificReports.9(1):17884.Bibcode:2019NatSR...917884M.doi:10.1038/s41598-019-54074-5.ISSN 2045-2322.PMC 6884510.PMID 31784547. ^"Facinggenderbiasinfacialrecognitiontechnology".HelpNetSecurity.August27,2020.RetrievedMay9,2021. ^Palmer,MatthewA.;Brewer,Neil;Horry,Ruth(March2013)."Understandinggenderbiasinfacerecognition:Effectsofdividedattentionatencoding".ActaPsychologica.142(3):362–369.doi:10.1016/j.actpsy.2013.01.009.ISSN 0001-6918.PMID 23422290.S2CID 205260206. ^"WhyGender-NeutralFacialRecognitionWillChangeHowWeLookatTechnology".InformaticsfromTechnologyNetworks.RetrievedMay9,2021. ^"FacialRecognition|GenderedInnovations".genderedinnovations.stanford.edu.RetrievedMay9,2021. ^Mason,SusanE.(September27,2007)."Ageandgenderasfactorsinfacialrecognitionandidentification".ExperimentalAgingResearch.12(3):151–154.doi:10.1080/03610738608259453.PMID 3830234. ^abFacialrecognitionsoftwarehasagenderproblem,retrievedMay9,2021 ^Meek,James(June13,2002)."Robocop".London:UKGuardiannewspaper. ^"BirminghamCityCentreCCTVInstallsVisionics'FaceIt".BusinessWire.June2,2008. ^Willing,Richard(September2,2003)."Airportanti-terrorsystemsflubtests;Face-recognitiontechnologyfailstoflag'suspects'".USAToday.Archivedfromtheoriginal(Abstract)onOctober1,2007.RetrievedSeptember17,2007. ^Meyer,Robinson(2015)."HowWorriedShouldWeBeAboutFacialRecognition?".TheAtlantic.RetrievedMarch2,2018. ^White,David;Dunn,JamesD.;Schmid,AlexandraC.;Kemp,RichardI.(October14,2015)."ErrorRatesinUsersofAutomaticFaceRecognitionSoftware".PLOSONE.10(10):e0139827.Bibcode:2015PLoSO..1039827W.doi:10.1371/journal.pone.0139827.PMC 4605725.PMID 26465631. ^"EFFSuesFBIForAccesstoFacial-RecognitionRecords".ElectronicFrontierFoundation.June26,2013. ^"Q&AOnFace-Recognition".AmericanCivilLibertiesUnion. ^abHarleyGeiger(December6,2011)."FacialRecognitionandPrivacy".CenterforDemocracy&Technology.RetrievedJanuary10,2012. ^abcCackley,AliciaPuente(July2015)."FACIALRECOGNITIONTECHNOLOGYCommercialUses,PrivacyIssues,andApplicableFederalLaw"(PDF). ^ThomasBrewster(September22,2020)."ThisRussianFacialRecognitionStartupPlansToTakeIts'AggressionDetection'TechGlobalWith$15MillionBackingFromSovereignWealthFunds".Forbes.RetrievedOctober4,2020. ^"Singel-Minded:AnatomyofaBacklash,orHowFacebookGotan'F'forFacialRecognition".WIRED.RetrievedApril10,2018. ^"FacebookCanNowFindYourFace,EvenWhenIt'sNotTagged".WIRED.RetrievedApril10,2018. ^"FacebookKeepsGettingSuedOverFace-RecognitionSoftware,AndPrivacyGroupsSayWeShouldBePayingMoreAttention".InternationalBusinessTimes.September3,2015.RetrievedApril5,2016. ^Herra,Dana."JudgetossesIllinoisprivacylawclassactionvsFacebookoverphototagging;Californiacasesstillpending".cookcountyrecord.com.RetrievedApril5,2016. ^Skinner-Thompson,Scott(2020).PrivacyattheMargins.CambridgeUniversityPress.p. 110.ISBN 9781107181373. ^Murgia,Madhumita(August12,2019)."London'sKing'sCrossusesfacialrecognitioninsecuritycameras".FinancialTimes(subscriptionsite).RetrievedAugust17,2019. ^"King'sCrossfacialrecognitioninvestigated".BBCNews.August15,2019.RetrievedAugust17,2019. ^Cellan-Jones,Rory(August16,2019)."TechTent:Isyourfaceonawatchlist?".BBCNews.RetrievedAugust17,2019. ^Sabbagh,Dan(September2,2019)."FacialrecognitiontechnologyscrappedatKing'sCrosssite".TheGuardian.ISSN 0261-3077.RetrievedSeptember2,2019. ^Sabbagh,Dan(October4,2019)."Facialrecognitionrow:policegaveKing'sCrossownerimagesofsevenpeople".TheGuardian.RetrievedOctober4,2020. ^abJudiciaryUK(August11,2020)."Judgement:BridgesvSouthWalesPolice-CourtsandTribunalsJudiciary"(PDF).JudiciaryUK.RetrievedSeptember10,2021. ^"PhotoAlgorithmsIDWhiteMenFine—BlackWomen,NotSoMuch".WIRED.RetrievedApril10,2018. ^JoyBuolamwini;TimnitGebru(2018)."GenderShades:IntersectionalAccuracyDisparitiesinCommercialGenderClassification".ProceedingsofMachineLearningResearch.Vol. 81.pp. 77–91.RetrievedMarch8,2018. ^Grother,Patrick;Quinn,George;Phillips,P.Jonathon(August24,2011)."ReportontheEvaluationof2DStill-ImageFaceRecognitionAlgorithms"(PDF).NationalInstituteofStandardsandTechnology. ^Buranyi,Stephen(August8,2017)."Riseoftheracistrobots–howAIislearningallourworstimpulses".TheGuardian.RetrievedApril10,2018. ^abcBrel,Ali(December4,2017)."Howwhiteengineersbuiltracistcode–andwhyit'sdangerousforblackpeople".TheGuardian.RetrievedApril10,2018. ^ab"FaceRecognitionVendorTest(FRVT)Ongoing".NIST.December14,2016.RetrievedFebruary15,2022. ^Grother,PatrickJ.;Ngan,MeiL.;Hanaoka,KayeeK.(December19,2019)."FaceRecognitionVendorTestPart3:DemographicEffects".{{citejournal}}:Citejournalrequires|journal=(help) ^RonaldLeenes;RosamundevanBrakel;SergeGutwirth;PauldeHert,eds.(2018).DataProtectionandPrivacy:TheInternetofBodies.BloomsburyPublishing.p. 176.ISBN 9781509926213. ^"UnlawfulUseofFacialRecognitionTechnologyLeadtoGDPRPenaltyinSweden".ComplianceJunction.September1,2019.RetrievedSeptember20,2019. ^Bock,Lisa(2020).IdentityManagementwithBiometrics:Explorethelatestinnovativesolutionstoprovidesecureidentificationandauthentication.PacktPublishing.p. 320.ISBN 9781839213212. ^Pascu,Luana(March16,2020)."CaliforniaresidentsfileclassactionagainstClearviewAIbiometricdatacollectioncitingCCPA".BiometricUpdate.com.RetrievedOctober25,2020. ^Burt,Chris(February24,2020)."CanadianPrivacyCommissionersinvestigateClearviewAI,developguidanceforpoliceuseofbiometrics".BiometricUpdate.com.RetrievedOctober25,2020. ^Conger,Kate;Fausset,Richard;Kovaleski,SergeF.(May14,2019)."SanFranciscoBansFacialRecognitionTechnology".TheNewYorkTimes.ISSN 0362-4331.RetrievedMarch26,2020. ^ab"SanFranciscoBansAgencyUseofFacialRecognitionTech".Wired.ISSN 1059-1028.RetrievedMarch26,2020. ^"SomervilleBansGovernmentUseOfFacialRecognitionTech".wbur.org.RetrievedMarch26,2020. ^ab"SomervilleCityCouncilpassesfacialrecognitionban–TheBostonGlobe".TheBostonGlobe.RetrievedMarch26,2020. ^Haskins,Caroline(July17,2019)."OaklandBecomesThirdU.S.CitytoBanFacialRecognition".Vice.RetrievedApril11,2020. ^Nkonde,Mutale(2019)."AutomatedAnti-Blackness:FacialRecognitioninBrooklyn,NewYork".KennedySchoolReview.pp. 30–26.ProQuest 2404400349–viaProQuest. ^"EUdropsideaoffacialrecognitionbaninpublicareas:paper".Reuters.January29,2020.RetrievedApril12,2020. ^"Facialrecognition:EUconsidersban".BBCNews.January17,2020.RetrievedApril12,2020. ^BostonmayorOKsbanonfacialrecognitiontech ^BostonmayorOKsbanonfacialrecognitiontech ^RachelMetz."PortlandpassesbroadestfacialrecognitionbanintheUS".CNN.RetrievedSeptember13,2020. ^"WestLafayetteCityCouncilapprovesbanonfacialrecognitiontechnology"[2] ^"HumanRightsGroupsCallOnTheUniversityofMiamiToBanFacialRecognition".Forbes.RetrievedOctober27,2020. ^"ReclaimYourFace:BanBiometricMassSurveillance!".ReclaimYourFace.RetrievedJune12,2021. ^Governorsignspoliceoverhaulintolaw ^Massachusettsisoneofthefirststatestocreaterulesaroundfacialrecognitionincriminalinvestigations. ^Gates,Kelly(2011).OurBiometricFuture:FacialRecognitionTechnologyandtheCultureofSurveillance.NYUPress.p. 156.ISBN 9780814732090. ^Gates,Kelly(2011).OurBiometricFuture:FacialRecognitionTechnologyandtheCultureofSurveillance.NYUPress.p. 161.ISBN 9780814732090. ^Konar,Amit;Chakraborty,Aruna(2015).EmotionRecognition:APatternAnalysisApproach.JohnWiley&Sons.p. 185.ISBN 9781118130667. ^Konar,Amit;Chakraborty,Aruna(2015).EmotionRecognition:APatternAnalysisApproach.JohnWiley&Sons.p. 186.ISBN 9781118130667. ^Konar,Amit;Chakraborty,Aruna(2015).EmotionRecognition:APatternAnalysisApproach.JohnWiley&Sons.p. 187.ISBN 9781118130667. ^"FacialRecognitionMarket–GlobalForecastto2021".DigitalJournal.December30,2016.RetrievedOctober17,2020. ^abFowler,Gary(October14,2019)."HowEmotionalAIIsCreatingPersonalizedCustomerExperiencesAndMakingASocialImpact".Frobes.RetrievedOctober17,2020. ^"FacialRecognitionMarket–GlobalForecastto2021".DigitalJournal.December30,2016.RetrievedOctober17,2020. ^"EurekaParkReturns"(Pressrelease).NationalScienceFoundation.January7,2013.RetrievedFebruary3,2013. ^abHarvey,Adam."CVDazzle:CamouflagefromFaceDetection".cvdazzle.com.RetrievedSeptember15,2017. ^Heilweil,Rebecca(July28,2020)."Maskscanfoolfacialrecognitionsystems,butthealgorithmsarelearningfast".www.vox.com.RetrievedJune30,2022. ^abMarks,Paul(2020)."BlockingFacialRecognition".cacm.acm.org.RetrievedJune30,2022. ^"TheseGoofy-LookingGlassesCouldMakeYouInvisibletoFacialRecognitionTechnology".Slate.January18,2013.RetrievedJanuary22,2013. ^Hongo,Jun."EyeglasseswithFaceUn-RecognitionFunctiontoDebutinJapan".TheWallStreetJournal.RetrievedFebruary9,2017. ^Osborne,Charlie."Privacyvisorwhichblocksfacialrecognitionsoftwaresetforpublicrelease".ZDNet.RetrievedFebruary9,2017. ^Stone,Maddie."TheseGlassesBlockFacialRecognitionTechnology".Gizmodo.RetrievedFebruary9,2017. ^"HowJapan'sPrivacyVisorfoolsface-recognitioncameras".PCWorld.RetrievedFebruary9,2017. ^Cox,Kate(April10,2020)."Someshirtshideyoufromcameras—butwillanyonewearthem?".ArsTechnica.RetrievedApril12,2020. ^Schreiber,Hope."Worriedaboutfacialrecognitiontechnology?Juggalomakeuppreventsinvoluntarysurveillance".RetrievedJuly18,2019. ^Vincent,James(July28,2020)."Facemasksarebreakingfacialrecognitionalgorithms,saysnewgovernmentstudy".TheVerge.RetrievedAugust27,2020. ^Hern,Alex(August21,2020)."Facemasksgivefacialrecognitionsoftwareanidentitycrisis".TheGuardian.ISSN 0261-3077.RetrievedAugust24,2020. ^"FawkesAI-UniversityofChicago".RetrievedJune30,2022. Furtherreading[edit] Farokhi,Sajad;Shamsuddin,SitiMariyam;Flusser,Jan;Sheikh,U.U;Khansari,Mohammad;Jafari-Khouzani,Kourosh(2014)."NearinfraredfacerecognitionbycombiningZernikemomentsandundecimateddiscretewavelettransform".DigitalSignalProcessing.31(1):13–27.doi:10.1016/j.dsp.2014.04.008. "TheFaceDetectionAlgorithmSettoRevolutionizeImageSearch"(Feb.2015),MITTechnologyReview Garvie,Clare;Bedoya,Alvaro;Frankle,Jonathan(October18,2016).PerpetualLineUp:UnregulatedPoliceFaceRecognitioninAmerica.CenteronPrivacy&TechnologyatGeorgetownLaw.RetrievedOctober22,2016. "FacialRecognitionSoftware'SoundsLikeScienceFiction,'butMayAffectHalfofAmericans".AsItHappens.CanadianBroadcastingCorporation.October20,2016.RetrievedOctober22,2016.InterviewwithAlvaroBedoya,executivedirectoroftheCenteronPrivacy&TechnologyatGeorgetownLawandco-authorofPerpetualLineUp:UnregulatedPoliceFaceRecognitioninAmerica. Externallinks[edit] MediarelatedtoFacialrecognitionsystematWikimediaCommons APhotometricStereoApproachtoFaceRecognition(master'sthesis).TheUniversityoftheWestofEngland,Bristol. vteDifferentiablecomputingGeneral Differentiableprogramming Informationgeometry Statisticalmanifold NeuralTuringmachine Differentiableneuralcomputer Automaticdifferentiation Neuromorphicengineering Cabletheory Patternrecognition Tensorcalculus Computationallearningtheory Concepts Gradientdescent SGD Clustering Regression Overfitting Adversary Attention Convolution Lossfunctions Backpropagation Normalization Activation Softmax Sigmoid Rectifier Regularization Datasets Augmentation Diffusion Autoregression Programminglanguages Python Julia Application Machinelearning Artificialneuralnetwork Deeplearning Scientificcomputing ArtificialIntelligence Hardware IPU TPU VPU Memristor SpiNNaker Softwarelibrary TensorFlow PyTorch Keras Theano ImplementationAudio–visual AlexNet WaveNet Humanimagesynthesis HWR OCR Speechsynthesis Speechrecognition Facialrecognition AlphaFold DALL-E Verbal Word2vec Transformer BERT LaMDA NMT ProjectDebater IBMWatson GPT-2 GPT-3 Decisional AlphaGo AlphaZero Q-learning SARSA OpenAIFive Self-drivingcar MuZero Actionselection Robotcontrol People AlexGraves IanGoodfellow YoshuaBengio GeoffreyHinton YannLeCun AndrewNg DemisHassabis DavidSilver Fei-FeiLi Organizations DeepMind OpenAI MITCSAIL Mila GoogleBrain MetaAI Architectures Recurrentneuralnetwork(RNN) Longshort-termmemory(LSTM) Gatedrecurrentunit(GRU) Echostatenetwork Multilayerperceptron(MLP) Autoencoder Variationalautoencoder(VAE) Generativeadversarialnetwork(GAN) Portals Computerprogramming Technology Category Artificialneuralnetworks Machinelearning Authoritycontrol:Nationallibraries Israel UnitedStates Retrievedfrom"https://en.wikipedia.org/w/index.php?title=Facial_recognition_system&oldid=1095806019" Categories:FacerecognitionAutomaticidentificationanddatacaptureMachinelearningtaskSurveillanceVideosurveillanceHiddencategories:CS1maint:usesauthorsparameterCS1Dutch-languagesources(nl)CS1errors:missingperiodicalArticleswithobsoleteinformationfromJune2022AllWikipediaarticlesinneedofupdatingArticleswithshortdescriptionShortdescriptionisdifferentfromWikidataUsemdydatesfromSeptember2021AllarticleswithunsourcedstatementsArticleswithunsourcedstatementsfromSeptember2020Articlescontainingpotentiallydatedstatementsfrom2004AllarticlescontainingpotentiallydatedstatementsArticlescontainingpotentiallydatedstatementsfrom2018ArticleswithunsourcedstatementsfromApril2020CommonscategorylinkfromWikidataArticleswithJ9UidentifiersArticleswithLCCNidentifiers Navigationmenu Personaltools NotloggedinTalkContributionsCreateaccountLogin Namespaces ArticleTalk English Views ReadEditViewhistory More Search Navigation MainpageContentsCurrenteventsRandomarticleAboutWikipediaContactusDonate Contribute HelpLearntoeditCommunityportalRecentchangesUploadfile Tools WhatlinkshereRelatedchangesUploadfileSpecialpagesPermanentlinkPageinformationCitethispageWikidataitem Print/export DownloadasPDFPrintableversion Inotherprojects WikimediaCommons Languages العربيةবাংলাCatalàČeštinaDanskEestiΕλληνικάEspañolفارسیFrançaisGalego한국어ՀայերենBahasaIndonesiaעבריתLatinaBahasaMelayuNederlands日本語NorskbokmålPortuguêsРусскийShqipСрпски/srpskiSuomiSvenskaதமிழ்ไทยTürkçeУкраїнськаTiếngViệt粵語中文 Editlinks