- Ukuqinisekiswa kobuso okusekelwe kusiphequluli okulula ngokuhlolwa kokuphila kusetshenziswa imodi yekhamera noma ukuqhathanisa izithombe ezimile ukuze kutholakale izimo ezinobungozi obuphansi.
- Ukuhlanganiswa okuguquguqukayo ngokusebenzisa ama-callback, imicimbi eyenziwe ngokwezifiso kanye ne-postMessage, ukusekela ukushumeka kwe-iframe kanye nokuxhumana phakathi kwamaphrojekthi.
- Imingcele elungisekayo yokuvula umlomo, ukujika ikhanda, imikhawulo yokwehluleka kanye nokuzinza okufanayo ukuze kulungiswe ukuphepha uma kuqhathaniswa nolwazi lomsebenzisi.
- Ifaneleka kakhulu izinhlelo zangaphakathi, ukuba khona, ukungena okulula kanye namacala okusetshenziswa kokufunda, hhayi amabhange aphephile kakhulu noma i-KYC kahulumeni.
Ukuqashelwa kobuso ku-inthanethi kushintshe kusuka ekubeni yindlela yobuqili emnandi yaba yindlela ewusizo yokuqinisekisa abasebenzisi, ukungena ngemvume kubantu noma ukuphatha ukungena ngaphandle kwehadiwe eyengeziwe noma izinhlelo zokusebenza zasendaweni. Iphakheji ye-npm evame ukubizwa ngokuthi “humanfacecheck” ifanelana ngqo nalesi simo ngokunikeza umsebenzi wokuqinisekisa ubuso osekelwe kusiphequluli osebenza ngqo ohlangothini lweklayenti, ugcine ulwazi lulula futhi luphendula ngenkathi usakunikeza izici ezithuthukisiwe njengokutholwa kokuphila kanye nokuhlanganiswa okuguquguqukayo phakathi kwamaphrojekthi.
Esikhundleni sokuthembela kumapayipi asindayo ohlangothini lweseva noma kuma-SDK ayinkimbinkimbi, lolu hlobo lwesisombululo lusebenzisa ubuchwepheshe obufana ne-face-api.js, i-TensorFlow.js kanye namamodeli amancane okuthola ubuso ukuze enze ukuqagela kwesikhathi sangempela kusiphequluli somsebenzisi. Lokho kusho ukuthi ungaqinisekisa ubuwena usebenzisa ikhamera noma izithombe ezinganyakazi, uzihlanganise nezinhlelo zokusebenza zewebhu ezikhona ngama-iframes kanye ne-postMessage, ulungise ukuziphatha ngamafayela okucushwa bese ukhetha phakathi kokugeleza okuphephile okusekelwe ekuphileni noma ukuqhathanisa izithombe okusheshayo nokungaphephile kuye ngezidingo zakho.
Lokho iphakheji ye-npm humanfacecheck eyenzelwe ukukwenza
Empeleni, iphakheji yesitayela se-npm humanfacecheck iyisistimu ekhanyayo yangaphambi kokuphela yokuqinisekisa ubuwena obusekelwe ebusweni obufaka ngqo ekhasini lewebhu noma kuhlelo lokusebenza lwewebhu. Isebenza ngokuphelele kusiphequluli, ngakho-ke azikho izingxenye zomdabu ezengeziwe ezidingekayo, futhi igxile kakhulu ekwenzeni umsebenzisi ageleze kahle ngenkathi isanikeza onjiniyela izikhonkwane zokulawula indlela ukuqinisekiswa okusebenza ngayo kanye nendlela imiphumela esetshenziswa ngayo.
Umgomo oyinhloko ukuqinisekisa ukuthi umuntu ophambi kwedivayisi ufana nesithombe sobuso esibhekiselwe kuye esebenzisa iseshini yekhamera ebukhoma noma izithombe ezinganyakazi. Ngaphezu kwalokho, isekela ukuhlolwa "kokuphila" kusetshenziswa izenzo ezilula njengokuvula umlomo noma ukujika ikhanda, okusiza ukuvimbela imizamo yokukhohlisa ngezithombe eziphrintiwe noma amavidiyo aqoshiwe kusengaphambili. Lokhu kuyenza ifaneleke kahle ukuhlolwa kobunikazi kwansuku zonke okubalulekile kodwa okungekho ezingeni elifanayo lengozi njengezinqubo ze-KYC zezinga lasebhange.
Ngokombono wokuhlanganisa, uhlelo lwakhiwe ukuze lusebenze kahle kumaphrojekthi namakhasi ahlukene, okuhlanganisa nokusethwa kwesizinda esihlukene. Ungayishumeka njenge-iframe, uxhumane nge-window.postMessage, futhi ulalele imicimbi noma ama-callback lapho ukuqinisekiswa kuqediwe. Lokhu kukuvumela ukuthi ugcine i-UI yokuqinisekisa kanye ne-logic zihlukanisiwe ngenkathi usaxhumanisa imiphumela nokugeleza kohlelo lwakho lokusebenza okuyinhloko njengokungena ngemvume, ukulandelela ukuba khona noma ukuvunyelwa kwangaphakathi kanye nokuqonda. ama-riesgos nama-controles.
Ngenxa yokuthi konke kusebenza kusiphequluli, ukusebenza kanye nokuphendula kubalulekile, futhi iphakheji igcinwa ilula ngamabomu ngokusebenzisa amamodeli asebenzayo kanye ne-UI ebalulekile kanye ne-logic kuphela. Ithembele emitatsheni yokufunda komshini ohlangothini lwamakhasimende kanye namamodeli okuthola ubuso alungiselelwe, ukuze ukwazi ukuyisebenzisa ekusingatheni iwebhu okuvamile ngaphandle kokudinga amaseva asekelwa yi-GPU noma ingqalasizinda ye-ML eyinkimbinkimbi.
Izici eziyinhloko: ukubhaliswa, ukuphila kanye nokuqinisekiswa okubukhoma
Isethi yezici zephakheji ye-npm yesitayela se-humanfacecheck igxile emjikelezweni wokuphila ophelele wokuqinisekiswa okusekelwe ebusweni: kusukela ekubhaliseni isithombe sokubhekisela kuya ekwenzeni ukuhlolwa okuqinile kwesikhathi sangempela. Esikhundleni sokunikeza i-API yokuqaphela engakahlelwanga, ihlanganisa konke okudingayo ukusekela ukugeleza kobunikazi obuvamile ezinhlelweni zokusebenza zewebhu.
Ukubhaliswa kobuso (ukubhaliswa) kuyibhlokhi yokuqala enkulu, ekuvumela ukuthi ubhalise ubuwena bomsebenzisi usebenzisa isithombe esilayishwe endaweni noma i-URL yesithombe esikude. Ngokulayishwa kwendawo, umsebenzisi ukhetha ifayela kusuka kudivayisi yakhe, bese licutshungulwa kusiphequluli. Ngokubhaliswa okusekelwe ku-URL, ukhomba uhlelo esithombeni esitholakala ku-inthanethi. Le ndlela ekabili ikunika ukuguquguquka uma usuvele unezithombe zephrofayela ezigciniwe noma uma ufuna ukuzithwebula zisanda kuphuma kukhamera yomsebenzisi.
Elinye lamakhono avelele ukubona ukuthi izinto ziphila kahle, okwengeza ungqimba olwengeziwe lokuvikela ekuqolweni. Esikhundleni sokuhlola nje ukuthi ubuso obubili bufana yini, uhlelo lucela umsebenzisi ukuthi enze izenzo ezithile, njengokuvula umlomo wakhe isikhashana noma ukuphendulela ikhanda ohlangothini olulodwa bese kuba kolunye. Lokhu kuhlola okusekelwe ekunyakazeni kusebenza kakhulu ekuhlungeni izithombe ezisicaba, izikrini noma ukuphinda ividiyo, ngoba kudinga ukusabela kwe-3D kwangempela okuvela kumuntu ophilayo.
Ngaphezu kokubhalisa kanye nokuphila, kunemodi yokuqinisekisa ngesikhathi sangempela lapho ikhamera yesiphequluli ithwebula khona ozimele bese iwaqhathanisa ngokuqhubekayo nethempulethi yokubhekisela. Njengoba umsebenzisi ehamba phambi kwekhamera, izici zobuso ziyatholwa, zikhishwe futhi zifaniswe nohlaka ngohlaka. Lapho uhlelo lufinyelela ukufana okuzinzile phezu kohlaka olulandelanayo, ukuqinisekiswa kubhekwa njengokuphumelelayo, futhi uhlelo lwakho lokusebenza lungaqhubeka nokungena ngemvume, ukungena noma yisiphi isenzo osinamathisela empumelelweni.
Ezimweni lapho ungakwazi noma ungathandi ukucela ukufinyelela ikhamera, iphakheji ifaka imodi yokuqhathanisa izithombe emsulwa encike ezithombeni ezinganyakazi esikhundleni sevidiyo ebukhoma. Kule modi, unikeza isithombe esibhekiselayo kanye nokuthwebula okusha, bese uhlelo luziqhathanisa ngaphandle kokwenza ukuhlolwa kokuphila. Lushintsha ukuphepha okuthile ukuze luhambisane namadivayisi avinjelwe noma abasebenzisi abaqaphela ubumfihlo abangafuni ukunikeza izimvume zekhamera.
Imodi yekhamera vs imodi yokuqhathanisa isithombe
Indlela ye-npm humanfacecheck ihlukanisa ngokucacile phakathi kokugeleza okusekelwe kwikhamera okuzenzakalelayo kanye nokugeleza kokuqhathanisa izithombe okungaguquki, ngayinye inezici zayo zokuphepha kanye namacala okusetshenziswa afanele. Ukuqonda umehluko phakathi kwalezi zinto ezimbili kukusiza ukuthi ukhethe imodi efanele kuye ngokuthi isimo sakho sibucayi kangakanani.
Kumodi yekhamera, isiphequluli sicela imvume yokusebenzisa ikhamera yomsebenzisi bese sisakaza ozimele bevidiyo bukhoma kuphayiphi yokubona nokuthola ubuso. Lokhu kwenza kube nokwenzeka ukuthola ukuthi uphila kahle ngoba uhlelo lungahlaziya ukunyakaza kanye namaphethini esikhathi, hhayi nje isithombe esisodwa. Ngokombono wokuphepha, lokhu kuyindlela enamandla kakhulu ngoba kwenza kube nzima kakhulu ngomhlaseli ukukhohlisa uhlelo esebenzisa izithombe ezilula noma amavidiyo aqoshiwe ngaphambilini aboniswe kwesinye isikrini.
Ngokuphambene nalokho, imodi yokuqhathanisa izithombe ayidingi ukufinyelela kwekhamera futhi isebenza ngokuqhathanisa izithombe ezimbili ezinganyakazi. Kokubili isithombe esibhekiselwe kuso kanye nesithombe esifanelekile kungalayishwa noma kunikezwe njengama-URL, futhi uhlelo luhlola kuphela ukuthi ubuso buyafana yini ngokwesilinganiso sokufana. Lokhu kulula, kuyashesha futhi kuvame ukuba lula ukukuhlanganisa ekugelezeni okuphansi kokungqubuzana, kodwa akunikezi isivikelo esinengqondo kumuntu ophethe isithombe sekhwalithi ephezulu somsebenzisi osemthethweni.
Imiphumela yokuphepha isobala: imodi yekhamera ibhekwa njengephephile ephezulu ngenxa yokutholakala kokuphila, kuyilapho imodi yokuqhathanisa izithombe ihlukaniswa ngamabomu njengokuphepha okuphansi. Ngenxa yalesi sizathu, inketho yesithombe kuphela ivame ukunconywa ezimweni ezinobungozi obuphansi lapho ubuthakathaka bomphumela ongewona oqondile bunqunyelwe, njengemiboniso ejabulisayo, izivivinyo zokuqeqesha noma amathuluzi angaphakathi angabalulekile. Ngokuphambene nalokho, noma yini ehilela idatha ebucayi, ukuthengiselana kwezezimali noma iziqinisekiso zobunikazi eziqinile kufanele ithembele ekuhlolweni kokuphila okusekelwe kukhamera noma ngisho nezixazululo ezithuthukisiwe kakhulu, ezihlolwe ngobuchwepheshe.
Ngokombono ongokoqobo, lokhu kuhlukaniswa kusiza futhi ngolwazi lomsebenzisi kanye nokuthobela imithetho, ngoba ungakhetha ukuthi uzocela nini ukufinyelela ikhamera nokuthi uzobuyela nini ekulayisheni okungaguquki. Abanye abasebenzisi noma izindawo ziqinile kakhulu ngezimvume, ngakho-ke ukuba nendlela engenakhamera kungavimbela ukungqubuzana, kodwa kubalulekile ukubiza leyo ndlela ngokucacile ku-UX yakho njengokuphepha okubuthakathaka ukuze ababambiqhaza baqonde ukuguquguquka.
Indlela imiphumela yokuqinisekisa ethunyelwa ngayo kuhlelo lwakho lokusebenza
Uma ukugeleza kokuqinisekisa sekuqediwe, uhlelo lwakho lokusebenza ludinga indlela ehlanzekile yokwamukela umphumela bese lusebenza kuwo, futhi ukwakheka kwesitayela se-humanfacecheck kunikeza iziteshi eziningi zokubuyisa ngasikhathi sinye. Lokhu kuphindaphinda kwenza ingxenye iguquguquke kuzo zonke izakhiwo ezahlukene kanye namazinga okuxhumanisa phakathi kwamamojula.
Indlela yokuqala yokuhlanganisa iwukusebenzisa imisebenzi yokubuyisa ucingo oyidlulisela ngesikhathi sokuqalisa, ngokuvamile into efana ne-onSuccess kanye ne-onFail. Uma i-logic yokuqinisekisa inquma ukuthi umsebenzisi udlulile noma wehlulekile ekuhlolweni, lokhu kubizwa kabusha kuyaqaliswa nganoma yikuphi ukulayisha okufanele, okukuvumela ukuthi uqondise kabusha umsebenzisi, ubuyekeze isimo, ubhale umcimbi wokuhlola noma ubonise imiyalezo. Lena iphethini eqondile esebenza kahle uma ufaka ingxenye ngqo kusuka kukhodi yakho eyinhloko yangaphambili.
Indlela yesibili, ehlukaniswe kakhulu isekelwe emcimbini: ingxenye ithumela umcimbi owenziwe ngokwezifiso, ovame ukubizwa ngokuthi i-faceVerifyResult, ezinye izingxenye zekhodi yakho ezingawulalela. Ngokunamathisela umlaleli womcimbi, ungasabela emiphumeleni ngaphandle kokuhlanganisa ngqo i-logic yebhizinisi lakho nezinto zangaphakathi zengxenye yokuqinisekisa. Lokhu kunengqondo uma wakha izakhiwo ze-modular lapho izingxenye ezahlukene ze-UI zidinga ukuphendula emphumeleni noma uma ufuna ukugcina iwijethi yokuqinisekisa ubuso izimele.
Isiteshi sesithathu sisekelwe ku-postMessage API, ewusizo kakhulu lapho i-UI yokuqinisekisa isebenza ngaphakathi kwe-iframe efakwe komunye umsuka noma iphrojekthi. Uma inqubo isiphelile, i-iframe ithumela umlayezo efasiteleni layo lomzali, elingakwazi ukuphatha idatha ngokufanele. Leli phethini lilungele ukuhlanganiswa kwamaphrojekthi amaningi lapho isikhombimsebenzisi sokuqinisekisa ubuso sisingathwa njengesevisi ephakathi, kodwa sisetshenziswa izinhlelo zokusebenza eziningi zamakhasimende ezahlukene ezingabelani nge-codebase efanayo.
Zonke izindlela ezintathu zingasebenza ngesikhathi esisodwa, ngakho-ke ukhululekile ukusebenzisa noma iyiphi efanelana kahle nendlela uhlelo lwakho lokusebenza oluhlelwe ngayo, noma uzihlanganisele nezinjongo zokuqapha nokulungisa amaphutha. Isibonelo, ungase uthembele kuma-callback ukuze ushayele i-UX yakho ngenkathi ubhala imicimbi ye-faceVerifyResult ukuze uthole ama-analytics noma uthola ukuxhumana kwe-postMessage kudeshibhodi yokusingatha elandelela izikhathi eziningi ezifakiwe.
Izinto okufanele uzicabangele uma udlulisa izithombe nge-URL noma i-base64
Ngisho noma iphakheji ilungiselelwe ukusebenza kahle kuklayenti, indlela ohlinzeka ngayo ngezithombe ekugelezeni kokuqinisekisa inomthelela obonakalayo ekuphenduleni kanye nesivinini esibonwayo. Indlela odlulisa ngayo izithombe zokubhekisela, ikakhulukazi, ingaletha ukubambezeleka okwengeziwe uma ingaphathwanga ngokucophelela.
Uma ubhalisa noma uqinisekisa ubuso usebenzisa ama-URL ezithombe, isiphequluli sidinga ukulanda isithombe ngaphambi kokuthi noma yikuphi ukutholwa noma ukukhishwa kwesici kuqale. Uma lawo ma-URL ekhomba amafayela amakhulu, amaseva akude anezikhathi zokuphendula ezisheshayo noma amanethiwekhi ane-latency ephezulu, abasebenzisi bangase babe nokulibaziseka ngaphambi kokuba isikhombikubona sokuqinisekisa siphendule. Lokhu kungabonakala ikakhulukazi ekuxhumaneni kwedatha yeselula noma ezifundeni ezine-bandwidth elinganiselwe.
Ukuze kuncishiswe lokhu kubambezeleka, isincomo esivamile ukuthumela idatha yesithombe ngqo kusetshenziswa izintambo ezifakwe ikhodi ye-base64 ezihlanganiswe ne-postMessage, ikakhulukazi uma usebenza kuma-iframes noma ezizindeni ezahlukene. Ngokufaka idatha yesithombe kumthwalo womlayezo, ugwema ama-HTTP hops engeziwe futhi unikeza ingxenye yokuqinisekisa ukufinyelela okusheshayo kumaphikseli ayidingayo. Lokhu kunganciphisa kakhulu isikhathi sokulinda futhi kwenze ukusebenza kuqageleke kalula ngoba ulawula ngqo ukuthi idatha idluliselwa nini futhi kanjani.
Le ndlela yokudlulisa ngqo ikhanga kakhulu uma i-backend yakho isivele inokufinyelela esithombeni sokubhekisela somsebenzisi futhi ingasicubungula kusengaphambili, isike noma isicindezele ngaphambi kokusithumela ekugcineni. Ungaqinisekisa ukuthi isithombe sinosayizi ofanele futhi silungiselelwe ukutholwa kobuso, ngaleyo ndlela ulondoloze i-bandwidth futhi usheshise ukuhlaziywa. Ngokuphambene nalokho, ukudlulisa ama-URL esithombe asindayo ngokunganaki kungaholela ekunciphiseni okungadingekile kanye nokuhlangenwe nakho komsebenzisi okungacolisisiwe kahle.
Sekukonke, ukunaka indlela ohambisa ngayo idatha yesithombe kusiphequluli—okungcono ukuncika ku-base64 kanye ne-postMessage ekusethweni okuyinkimbinkimbi—kusiza ukugcina ukuhamba komsebenzi we-humanfacecheck kushesha futhi kulula ukukusebenzisa, okubalulekile ekusetshenzisweni kwezinhlelo zokusebenza zangempela.
Izinketho zokucushwa kokuphila nokuqina
Isixazululo sesitayela se-npm humanfacecheck siveza isethi yamapharamitha wokucushwa acwebezelayo, avame ukufakwa phakathi kufayela elifana ne-js/modules/config.js, okukunikeza ukulawula ukuthi ukutholwa kokuphila kanye ne-logic yokuqinisekisa kufanele kube okuqinile futhi okuphendulayo kangakanani. Ukulungisa la manani kukuvumela ukuthi ulungise ibhalansi phakathi kokuphepha, ukubekezelelana kokunyakaza komsebenzisi kanye nolwazi lomsebenzisi jikelele.
Ukulungiswa okukodwa okuyisihluthulelo yi-mouthOpenThreshold, evame ukuba cishe ku-0.7 ngokuzenzakalelayo, okunquma ukuthi umsebenzisi kufanele avule umlomo wakhe ububanzi obungakanani ukuze isenzo sibhekwe njengesisebenzayo. Umkhawulo ophezulu usho ukuthi uhlelo ludinga ukuvula umlomo okugqamile, okwenza kube nzima ukudlula ukuhlolwa ngengozi kodwa futhi kungase kube nzima kakhulu kubasebenzisi. Ngokuphambene nalokho, ukwehlisa umkhawulo kungenza umsebenzi ube lula kodwa kungase kunciphise kancane ukuzethemba kokuthi isenzo sihlosiwe.
Isethingi ye-mouthOpenDuration, enokuzenzakalelayo okufana nama-millisecond angu-800, ilawula ukuthi umlomo udinga ukuhlala isikhathi esingakanani uvulekile ukuze isenzo sokuphila sibalwe. Lesi sidingo sesikhathi sisiza ukuqinisekisa ukuthi uhlelo alubangelwa ukuvezwa okufushane, okungahleliwe. Ukwandisa isikhathi kungathuthukisa ukuqina ngokumelene nemizamo yokukhohlisa esheshayo, kuyilapho ukufinyeza kwenza ukugeleza kuzizwe kushesha futhi kukhululekile kubasebenzisi, ikakhulukazi labo abanezidingo zokufinyeleleka noma ukusabela okuhamba kancane.
Imingcele yokuhamba kwekhanda nayo ingalungiselelwa, ngokuvamile ichazwa ngokwehlukana ukuze kuphendulelwe ikhanda ngakwesokudla nakwesobunxele. Isibonelo, ungase ubone i-headShakeThreshold.ngqo cishe ku-1.5 kanye ne-headShakeThreshold.left eduze kuka-0.67. Amanani amakhulu abonisa ukuthi uhlelo lulindele ukujikeleza okukhulu kulolo hlangothi ngaphambi kokuphatha isenzo njengesisebenzayo, kuyilapho amanani amancane eqinisa ukubekezelelana futhi edinga ukunyakaza okukhulu. Ngenxa yokuthi abantu abahambi njalo ngokulinganayo, ukuba nezilungiselelo ezihlukene zesobunxele nesokudla kukuvumela ukuthi ulinganisele ukuziphatha kwemvelo kakhulu kubasebenzisi abahlukahlukene.
Ngale kokuthinta kokuphila, amapharamitha afana ne-maxFailCount kanye ne-requiredMatchFrames alawula ukuthi inqubo yokuqinisekisa ithethelela futhi izinze kangakanani. Okuzenzakalelayo kwe-maxFailCount okungaba ngu-4 kubonisa ukuthi zingaki imizamo ehlulekile elandelanayo ebekezelelwa ngaphambi kokuba uhlelo luyeke futhi lubike ukwehluleka okuphelele, okusiza ukugwema ukuzama okungapheli kanye nokuhlola okungenzeka kube yi-brute-force. Isethingi ye-requiredMatchFrames, evame ukuba ku-3 ngokuzenzakalelayo, icacisa ukuthi zingaki izinhlaka zevidiyo ezilandelanayo okumele zibonise ukufana okuphumelelayo ngaphambi kokuba uhlelo luqinisekise ubunikazi, okuhlunga ama-blips okuthola okwesikhashana futhi kwenze umphumela ube nokwethenjelwa kakhudlwana.
Ngokulungisa ngokucophelela lezi zinketho zokucushwa, ungenza ukuziphatha kwe-humanfacecheck kuhambisane nomongo wohlelo lwakho lokusebenza—kungakhathaliseki ukuthi ukhetha ukuphepha okuqinile kokuqinisekiswa kwabasebenzi bangaphakathi noma ukugeleza okukhululekile kokungena okuvamile kanye nama-demo.
Amacala okusetshenziswa ajwayelekile kanye nalapho okungafanele kusetshenziswe khona
Umklamo wephakheji ye-npm yesitayela se-humanfacecheck uhlose ngokucacile izimo zokusetshenziswa kwansuku zonke, ezingokoqobo kunezimo zezimali noma zomthetho ezibucayi kakhulu. Lokho kwenza kube kuhle kakhulu emisebenzini eminingi esekwe kuwebhu lapho ukulula kubalulekile khona futhi iphrofayili yengozi iphakathi nendawo.
Isicelo esisodwa esivamile ukuqinisekiswa kobunikazi bangaphakathi ezinhlelweni zezinkampani noma zenhlangano. Isibonelo, abasebenzi bangasebenzisa ukuqinisekiswa kobuso ukufinyelela amadeshibhodi angaphakathi, bavume izenzo ezingezona ezibucayi noma baqinisekise ukuba khona kwabo lapho beqala ukushintsha. Ngenxa yokuthi indawo ilawulwa kancane futhi ngokuvamile kunezendlalelo zokuphepha ezengeziwe (njenge-VPN noma izimvume ezisekelwe endimeni), le ndlela yokuqinisekisa ingeza isiqinisekiso esingenazinkinga ngaphandle kokudinga izinqubo ezisindayo ze-KYC.
Esinye isimo esidumile yizimo zokuba khona noma zokungena, lapho ufuna ukuqinisekisa ukuthi umuntu othile ukhona ngokoqobo endaweni ethile noma uhlanganyela emsebenzini othile. Cabanga ngamahhovisi, izindawo zokusebenza ndawonye, izikhathi zokuqeqesha, izingqungquthela noma amakilasi lapho ukuqinisekiswa kobuso kuthatha indawo noma kwengeza amashidi okungena ngesandla noma ukuswayipha amabheji. Ukuhlolwa kokuphila okusekelwe kukhamera kusebenza kahle kakhulu lapha ngoba kungaqinisekisa ngokushesha ukuba khona ngaphandle kwehadiwe eyinkimbinkimbi.
Izinhlelo zokusebenza zabathengi nazo zingazuza ekuqinisekisweni okunjalo, ikakhulukazi kokungena ngemvume okulula kohlelo lokusebenza okungabandakanyi izibambiso ezinkulu zezimali noma iziqinisekiso zobunikazi ezisemthethweni. Abasebenzisi bangangena ngemvume kuhlelo lokusebenza lwewebhu noma lwe-hybrid besebenzisa ubuso babo esikhundleni sokuthayipha amaphasiwedi njalo, okuthuthukisa ukusebenziseka kalula ngenkathi besanikeza ukungqubuzana okungcono kunepheya elilula lomsebenzisi nephasiwedi. Kulezi zimo, ukuhlanganisa ukuqinisekiswa kobuso nezinye izici ezifana nokuqinisekiswa kwe-imeyili noma ukuqashelwa kwedivayisi kungaletha ukuphepha okuqinile ngaphandle kokufinyelela izinga eliphelele lebhizinisi.
Izindawo zemfundo, ama-demo kanye namaphrojekthi okufunda nawo afanelekile: abafundi noma abathuthukisi bangazama ukubona ubuso kanye nemibono yokuphila endaweni esekelwe kusiphequluli ngaphandle kokutshala imali engqalasizinda eyinkimbinkimbi. Lokhu kungasetshenziselwa ukufundisa imiqondo yokufunda komshini, ukwenza iphrothoyiphu yokugeleza okusha kwe-UX noma ukubonisa amakhono okubona ikhompyutha emicimbini kanye nama-hackathon.
Kodwa-ke, kubalulekile ukungasebenzisi lolu hlobo lokuqinisekiswa kobuso okulula, okuseceleni kwamakhasimende njengendlela eyinhloko yokuqinisekisa ubuwena ezimweni zokuphepha okuphezulu njengokuvulwa kwama-akhawunti asebhange, ukuqinisekiswa kobuwena ezingeni likahulumeni noma ukungeniswa kwemithetho eqinile. Lezo zimo zidinga izinhlelo eziqinile, ezihloliwe ezivame ukusekelwa abahlinzeki bamafu abakhethekile, ukuhlolwa kwezinto eziningi, ukuqinisekiswa kwedokhumenti, ukuqapha ukukhwabanisa kanye nokuthobela imithetho eqinile. Isixazululo esisekelwe kusiphequluli esichazwe lapha asihlose ukufaka esikhundleni salezo; siyanezela ezimweni zokusetshenziswa okuphansi lapho isivinini kanye nolwazi lomsebenzisi kubaluleke kakhulu kunezinga eliphezulu lokuqinisekisa.
Ubuchwepheshe obuyisisekelo kanye nokukhetha kwamamodeli
Ngaphansi kwe-hood, iphakheji ye-npm yesitayela se-humanfacecheck ivame ukuthembela ekuhlanganisweni kwemitapo yolwazi yokufunda komshini yesimanje yeJavaScript kanye namamodeli enethiwekhi ye-neural acwebezelayo enzelwe isiphequluli. Lesi sitaki senza kube lula ukubona ubuso okuqinile nokubonwa ngaphandle kokuphazamisa uhlaka ngalunye kuseva ekude.
Ingxenye eyinhloko yephazili yi-face-api.js, umtapo wolwazi odumile wezinga eliphezulu owakhelwe phezu kwe-TensorFlow.js ohlinzeka ngamamodeli aqeqeshwe kusengaphambili okuthola ubuso, indawo yezimpawu kanye nokushumeka izici. Nge-face-api.js, uhlelo lungathola ubuso kufreyimu ngayinye yevidiyo, lukhiphe amaphuzu abalulekile ebusweni (njengamehlo, impumulo, namakhona omlomo) bese lubala amavekhtha ezincazelo amelela izici ezihlukile zobuso. Lezi zincazelo zingaqhathaniswa namathempulethi abhalisiwe ukuze kunqunywe ukuthi ubuso obubili bungobomuntu oyedwa.
I-TensorFlow.js isebenza njengesikhathi sokusebenza esisebenzisa lawa manethiwekhi e-neural ngqo kusiphequluli kusetshenziswa i-WebGL nezinye izindlela zokusheshisa. Ilayisha izisindo zemodeli, yenza ama-convolutions kanye neminye imisebenzi, futhi ibuyise imiphumela ngesivinini esisebenzisanayo. Ngenxa yokuthi isebenza ngokuphelele kuklayenti, le ndlela igcina idatha ye-biometric kudivayisi yomsebenzisi ngesikhathi sokucabanga, inciphisa ukusetshenziswa kwe-bandwidth futhi ikunikeze ukulawula okwengeziwe kokugeleza kwedatha.
Ukuze kugcinwe iphakheji ilula, kusetshenziswa ama-detector anesitayela sobuso obuncane njenge-TinyFaceDetector ukuze kutholakale indawo yobuso ekuqaleni. Lawa mamodeli alungiselelwe ngokukhethekile isivinini kanye nenkumbulo, ehweba ngokunemba okuphelele kokusebenza kwesikhathi sangempela kumadivayisi ahlukahlukene, kufaka phakathi ama-laptop amadala nama-smartphone aphakathi nendawo. Ezimweni eziningi zokusetshenziswa kokuqinisekisa lapho umsebenzisi eseduze nekhamera, ama-detector anjalo anele.
Ngokuhlanganisa lobu buchwepheshe, iphakheji ye-npm inganikeza ipayipi lokuqinisekisa elisekelwe kusiphequluli elizizwa lisabela ngenkathi liletha imiphumela enenjongo, konke ngaphansi kwelayisensi evumelekile njenge-MIT ekhuthaza ukuzama nokuhlanganiswa kumaphrojekthi ezentengiselwano nawomthombo ovulekile ngokufanayo.
Sekukonke, lesi sigaba sobuchwepheshe sibonisa ukuthi ukufunda komshini okungaphakathi kwesiphequluli sekuhambe ibanga elingakanani, okwenza kube nokwenzeka ukusebenzisa ukuqinisekiswa kobuso kanye nokugeleza kokuphila ngokuphelele ku-JavaScript ngaphandle kokuthembela okukhulu kwemvelo.
Ukuhlanganisa konke, iphakheji ye-npm yesitayela se-humanfacecheck inikeza ulwazi lokuqinisekisa ubuso besiphequluli oluhlanganisa ukuhlanganiswa okulula kwe-front-end, ukuhlolwa kokuphila okulungisekayo, izindlela eziningi zokulethwa kwemiphumela kanye nomehluko ocacile phakathi kokugeleza okusekelwe kukhamera okuphephile kanye nokuqhathaniswa okulula kwezithombe ezimile. Uma isetshenziswa ezimweni ezifanele—njengezinhlelo zangaphakathi, ukulandelela ukuba khona, ukungena ngemvume kwansuku zonke kohlelo lokusebenza kanye nama-demo ezemfundo—inikeza ibhalansi esebenzayo yokulula nokuvikeleka, ngenkathi isashiya isikhala sezinsizakalo zamafu eziqinile, ezisezingeni lobuchwepheshe noma nini lapho udinga ukusingatha ukuqinisekiswa kobunikazi obuyingozi kakhulu.
