Tekinoroji uye mamodheru eiyo ramangwana rekuona komputa system yakagadzirwa uye yakagadziridzwa zvishoma nezvishoma uye mumapurojekiti akasiyana ekambani yedu - muMail, Cloud, Tsvaga. Vakakura sechizi yakanaka kana cognac. Rimwe zuva takaona kuti edu neural network anoratidza mhedzisiro yakanaka mukuzivikanwa, uye isu takasarudza kuvasanganisa kuita chimwe chete b2b chigadzirwa - Vision - iyo yatinoshandisa isu pachedu uye tinokupa iwe kuti ushandise.

Nhasi, tekinoroji yedu yekuona komputa paMail.Ru Cloud Solutions papuratifomu iri kubudirira kushanda nekugadzirisa matambudziko akaomarara anoshanda. Izvo zvakavakirwa pane akati wandei neural network akadzidziswa pamaseti edu e data uye nyanzvi mukugadzirisa matambudziko akaiswa. ese masevhisi anomhanya pane yedu server zvivakwa. Iwe unogona kubatanidza iyo yeruzhinji Vision API mumashandisirwo ako, kuburikidza ayo ese ekuita sevhisi anowanikwa. Iyo API inokurumidza - nekuda kweseva GPUs, avhareji nguva yekupindura mukati metiweki yedu i100 ms.

Enda kukati, pane nyaya yakadzama uye mienzaniso yakawanda yebasa reVision.

Muenzaniso wesevhisi umo isu pachedu tinoshandisa yakataurwa matekinoroji ekuzivikanwa kwechiso ndeye Events. Chimwe chezvikamu zvayo ndeye Vision photo stands, iyo yatinoisa pamisangano yakasiyana-siyana. Kana iwe ukasvika panzvimbo yakadai yepikicha, tora foto ine yakavakirwa-mukati kamera uye isa email yako, iyo sisitimu ichakurumidza kuwana pakati pemhando yemifananidzo iyo yawakatorwa nevashandi vemifananidzo vemusangano, uye, kana uchida, ichakutumira mafoto akawanikwa kwauri neemail. Uye isu hatisi kutaura nezve akatemerwa mapikicha emifananidzo - Vision inokuziva iwe kunyangwe kumashure chaiko mune chaunga chevashanyi. Ehe, haisi iyo foto inomira pachayo inozivikanwa, aya angori mahwendefa ari munzvimbo dzakanaka anongotora mafoto evaenzi nemakamera avo akavakirwa-mukati uye kutumira ruzivo kumaseva, uko mashiripiti ese ekuzivikanwa anoitika. Uye isu takaona kanopfuura kamwe chete kuti zvinoshamisa sei kubudirira kwehunyanzvi kunyange pakati penyanzvi dzekuziva mifananidzo. Pazasi isu tichataura nezvemimwe mienzaniso.

1. Yedu Face Recognition modhi

1.1. Neural network uye kumhanya kwekugadzirisa

Kuti tizivikanwe, tinoshandisa shanduko yeResNet 101 neural network model. Zvisinei, saizi yevector inomiririra 128, kwete 512. Seti yedu yekudzidziswa ine mapikicha anosvika mamirioni gumi evanhu 10.

Iyo modhi inomhanya nekukurumidza nekuda kweyakanyatsosarudzwa sevha yekumisikidza architecture uye GPU komputa. Zvinotora kubva ku100 ms kugamuchira mhinduro kubva kuAPI pane yedu yemukati network - izvi zvinosanganisira kuona chiso (kuona chiso mumufananidzo), kuziva uye kudzorera PersonID mumhinduro yeAPI. Nemavhoriyamu makuru e data rinouya - mapikicha nemavhidhiyo - zvinotora yakawanda nguva kuendesa iyo data kushumiro uye kugamuchira mhinduro.

1.2. Kuongorora kushanda kwemuenzaniso

Asi kuona kugona kweneural network ibasa rakaoma kunzwisisa. Hunhu hwebasa ravo hunoenderana nekuti ndeapi data seti mamodheru akadzidziswa uye kuti akagadziridzwa kushanda nedata chairo.

Takatanga kuongorora huchokwadi hwemodhi yedu neiyo yakakurumbira LFW verification bvunzo, asi idiki uye iri nyore. Mushure mekusvika 99,8% kunyatsoita, haichabatsiri. Pane makwikwi akanaka ekuongorora mamodheru ekuzivikanwa - Megaface, yatakasvika zvishoma nezvishoma 82% chinzvimbo 1. Muedzo weMegaface une miriyoni yemifananidzo - zvinokanganisa - uye modhi yacho inofanirwa kukwanisa kusiyanisa zviuru zvinoverengeka zvemifananidzo yevanozivikanwa kubva kuFacescrub. dataset kubva kune ditractors. Nekudaro, tabvisa iyo Megaface bvunzo yezvikanganiso, takaona kuti neyakacheneswa vhezheni tinowana huchokwadi hwe98% chinzvimbo 1 (mifananidzo yevanozivikanwa kazhinji yakanyatso kujeka). Nokudaro, vakagadzira muedzo wekuzivikanwa wakasiyana, wakafanana neMegaface, asi nemifananidzo ye "vanhuwo zvavo". Takabva tavandudza kurongeka kwekuziva pamadatasets edu ndokuenderera mberi. Mukuwedzera, isu tinoshandisa clustering quality bvunzo iyo ine zviuru zvakati wandei mafoto; inotevedzera tagging yechiso mugore remushandisi. Muchiitiko ichi, mapoka mapoka evanhu vakafanana, boka rimwe remunhu mumwe nomumwe anozivikanwa. Takaongorora hutano hwebasa pamapoka chaiwo (chokwadi).

Zvechokwadi, kukanganisa kwekuzivikanwa kunoitika nemhando ipi zvayo. Asi mamiriro ezvinhu akadaro anowanzogadziriswa nekugadzirisa zvakanaka zvikumbaridzo zvemamiriro ezvinhu chaiwo (kune misangano yose tinoshandisa zvipingamupinyi zvakafanana, asi, semuenzaniso, nokuda kwekugadzirisa zvirongwa zvekuwana tinofanira kuwedzera zvakanyanya zvikumbaridzo kuitira kuti pave nemaitiro mashoma enhema). Ruzhinji rwevashanyi vemusangano vakazivikanwa nenzira kwayo neVision yedu mafoto mabhomba. Dzimwe nguva mumwe munhu aitarisa zvakadzvanywa obva ati, "Sitimu yako yakakanganisa, handisi ini." Takabva tavhura pikicha yacho yese, zvikaonekwa kuti zvechokwadi paive nemuenzi uyu pamufananidzo, isu chete takanga tisiri kumutora mifananidzo, asi mumwe munhu, munhu wacho aingove ari kumashure mublur zone. Uyezve, iyo neural network inowanzo ziva nemazvo kunyangwe chikamu chechiso chisingaoneki, kana munhu akamira muprofile, kana kunyange hafu-yakatendeuka. Iyo sisitimu inogona kuziva munhu kunyangwe chiso chiri munzvimbo yekukanganiswa kwemaziso, toti, kana uchipfura neakafara-ekona lens.

1.3. Mienzaniso yekuedza mumamiriro ezvinhu akaoma

Pazasi pane mienzaniso yekuti yedu neural network inoshanda sei. Mapikicha anoendeswa kune yekuisa, iyo yaanofanira kunyora achishandisa PersonID - yakasarudzika identifier yemunhu. Kana mifananidzo miviri kana kupfuura iine ID yakafanana, saka, maererano nemhando, mafoto aya anoratidza munhu mumwe chete.

Ngationei nekukasira kuti kana tichiyedzwa, isu tinokwanisa kuwana akasiyana ma paramita uye modhi maburi atinogadzirisa kuti tiwane imwe mhedzisiro. Iyo yeruzhinji API yakagadziridzwa kuti inyatso kunyatsoenderana pane zvakajairika kesi.

Ngatitangei nechinhu chakareruka, nekuzivikanwa kwechiso chakatarisana.

Zvakanaka, izvo zvaive nyore. Ngatiomese basa, wedzera ndebvu uye mashoma emakore.

Vamwe vachataura kuti izvi zvakanga zvisinawo kuoma, nokuti muzviitiko zvose chiso chose chinooneka, uye ruzivo rwakawanda pamusoro pechiso chinowanikwa kune algorithm. Zvakanaka, ngatishandure Tom Hardy kuita chimiro. Dambudziko iri rakanyanya kuoma, uye takashandisa simba rakawanda kuti tibudirire kurigadzirisa tichichengetedza yakaderera mwero wekukanganisa: isu takasarudza seti yekudzidzira, yakafungwa kuburikidza nekuvaka kweiyo neural network, kukudza mabasa ekurasikirwa uye nekuvandudza pre-processing. yemifananidzo.

Ngatimupfekedze musoro.

Nenzira, iyi ndiyo muenzaniso wemamiriro ezvinhu akaoma zvikuru, sezvo chiso chakanyanyisa kuvharwa, uye mumufananidzo wezasi panewo mumvuri wakadzika unovanza maziso. Muhupenyu chaihwo, vanhu vanowanzochinja chitarisiko chavo nerubatsiro rwemagirazi erima. Ngatiite zvimwe chete naTom.

Zvakanaka, ngatiedzei kukanda mapikicha kubva kumazera akasiyana, uye panguva ino tichaedza nemumwe mutambi. Ngatitorei muenzaniso wakaoma zvikuru, apo shanduko dzine chekuita nezera dzinonyanya kutaurwa. Mamiriro acho haasi kure-kure; zvinowanzoitika kana iwe uchida kuenzanisa mufananidzo mupasipoti nechiso chemutakuri. Mushure mezvose, mufananidzo wekutanga unowedzerwa kune pasipoti kana muridzi ane makore makumi maviri, uye nezera re20 munhu anogona kuchinja zvikuru:

Iwe unofunga kuti nyanzvi huru pane isingagoneki mishoni haina kuchinja zvakanyanya nezera? Ndinofunga kuti kunyange vanhu vashomanana vaizobatanidza mapikicha epamusoro nepamusoro, mukomana akachinja zvakanyanya mumakore.

Neural network inosangana nekuchinja muchitarisiko kakawanda. Somuenzaniso, dzimwe nguva vakadzi vanogona kuchinja zvikuru mufananidzo wavo nerubatsiro rwezvizoro:

Zvino ngatiomese basa zvakanyanya: ngatiti zvikamu zvakasiyana zvechiso zvakafukidzwa nemifananidzo yakasiyana. Mumamiriro ezvinhu akadaro, algorithm haigoni kuenzanisa sampuli dzose. Nekudaro, Vision inobata mamiriro akadai zvakanaka.

Nenzira, panogona kunge paine zviso zvakawanda mumufananidzo; semuenzaniso, vanhu vanopfuura zana vanogona kukwana mumufananidzo wakajairwa wehoro. Aya mamiriro ezvinhu akaoma kune neural network, sezvo zviso zvakawanda zvichigona kuvhenekerwa zvakasiyana, zvimwe zvisiri zvekutarisa. Nekudaro, kana iyo foto ikatorwa neyakakwana resolution uye mhando (angangoita makumi manomwe neshanu pixels paskweya yakavhara kumeso), Vision inokwanisa kuiona nekuiziva.

Iyo yakasarudzika yemafoto ekutaura nemifananidzo kubva kumakamera ekutarisisa ndeyekuti vanhu vanowanzo sviba nekuti vanga vasingatarise kana kuti vaifamba panguva iyoyo:

Zvakare, kusimba kwemwenje kunogona kusiyanisa zvakanyanya kubva pamufananidzo kuenda kune mufananidzo. Izvi, zvakare, zvinowanzova chigumbuso; akawanda algorithms ane dambudziko rakakura kugadzira mifananidzo yakasviba zvakanyanya uye yakareruka, tisingarevi nekunyatso kuifananidza. Rega ndikuyeuchidze kuti kuti uwane mhedzisiro iyi unofanirwa kugadzirisa zvikumbaridzo neimwe nzira; chimiro ichi hachisati chawanikwa pachena. Isu tinoshandisa yakafanana neural network kune vese vatengi; ine zvikumbaridzo zvakafanira kune akawanda anoshanda mabasa.

Isu nguva pfupi yadarika takaburitsa vhezheni nyowani yemodhi inoona zviso zveAsia zvine hunyoro hwepamusoro. Iri raimbove dambudziko guru, iro raitonzi "muchina kudzidza" (kana "neural network") rusaruraganda. European neAmerica neural network yakaziva zviso zveCaucasian zvakanaka, asi neMongoloid neNegroid vakatarisana mamiriro acho aive akanyanya kuipa. Zvichida, muChina mamiriro acho akanga akasiyana chaizvo. Izvo zvese nezve kudzidzisa data seti dzinoratidza marudzi makuru evanhu mune imwe nyika. Zvisinei, mamiriro ezvinhu ari kuchinja; nhasi dambudziko iri harina kunyanya kuoma. Vision haina dambudziko nevanhu vemarudzi akasiyana.

Kuzivikanwa kwechiso chingori chimwe chezvakawanda zvekushandisa tekinoroji yedu; Vision inogona kudzidziswa kuziva chero chinhu. Semuyenzaniso, rezinesi mahwendefa, kusanganisira mumamiriro ezvinhu akaoma kune algorithms: pamakona akapinza, akasviba uye akaoma kuverenga rezinesi mahwendefa.

2. Zviitiko zvekushandisa zvinoshanda

2.1. Physical access control: kana vanhu vaviri vachishandisa pass yakafanana

Nerubatsiro rweVision, unogona kuita masisitimu ekurekodha kusvika uye kubuda kwevashandi. Iyo yechinyakare sisitimu yakavakirwa pamagetsi inopfuura ine zvipingamupinyi zviri pachena, semuenzaniso, unogona kupfuudza vanhu vaviri vachishandisa bheji rimwe. Kana iyo access control system (ACS) ichiwedzerwa neVision, icharekodha nekuvimbika kuti ndiani akauya / akasiya uye rinhi.

2.2. Nguva yekutevera

Iyi Vision yekushandisa kesi ine hukama neyakapfuura. Kana iwe ukawedzera iyo yekuwana sisitimu nebasa redu rekuzivikanwa kwechiso, haizongokwanisa kuona kukanganisa kwekutonga, asiwo kunyoresa kuvepo chaiko kwevashandi muchivako kana nzvimbo. Mune mamwe mazwi, Vision ichakubatsira iwe kunyatsofunga kuti ndiani akauya kubasa uye akasiya panguva ipi, uye ndiani akasvetuka basa zvachose, kunyangwe vamwe vake vakamuvhara pamberi pevakuru vake.

2.3. Vhidhiyo Yekuongorora: Vanhu Kuteedzera uye Chengetedzo

Nekutevera vanhu uchishandisa Vision, unogona kunyatso ongorora iyo chaiyo traffic yenzvimbo dzekutenga, zviteshi zvechitima, nzira, migwagwa nedzimwe nzvimbo dzakawanda dzeveruzhinji. Kutevera kwedu kunogonawo kubatsira zvakanyanya mukudzora kupinda, semuenzaniso, kunzvimbo yekuchengetera zvinhu kana imwe nzvimbo yakakosha yehofisi. Uye zvechokwadi, kutevera vanhu uye zviso kunobatsira kugadzirisa matambudziko ekuchengetedza. Wabata munhu arikuba muchitoro chako? Wedzera PersonID yake, iyo yakadzoserwa neVision, kune yakasviba yevhidhiyo analytics software yako, uye nguva inotevera sisitimu ichakurumidza kuzivisa kuchengetedzeka kana rudzi urwu rukaoneka zvakare.

2.4. Mukutengeserana

Retail uye akasiyana mabhizinesi ebasa anofarira kuzivikanwa kwemutsara. Nerubatsiro rweChiratidzo, unogona kuona kuti iyi haisi mhomho yevanhu, asi mutsara, uye uone kureba kwayo. Uye ipapo sisitimu inozivisa avo vanotungamira nezve mutsara kuti vaone mamiriro ezvinhu: kungave kune kuwanda kwevashanyi uye vamwe vashandi vanofanirwa kudanwa, kana mumwe munhu ari kunonoka pabasa ravo.

Rimwe basa rinonakidza nderokuparadzanisa vashandi vekambani muhoro kubva kune vashanyi. Kazhinji, sisitimu inodzidziswa kupatsanura zvinhu zvakapfeka dzimwe mbatya (yekupfeka kodhi) kana neimwe yakasarudzika chimiro (yakadhindwa sikavha, bheji pachipfuva, zvichingodaro). Izvi zvinobatsira kunyatsoongorora kuvapo (kuitira kuti vashandi vasa "kurisa" nhamba dzevanhu vari muhoro nekungovapo kwavo).

Uchishandisa kucherechedzwa kwechiso, iwe unogona zvakare kuongorora vateereri vako: chii kuvimbika kwevashanyi, ndiko kuti, vangani vanhu vanodzokera kunzvimbo yako uye neakawanda sei. Verenga kuti vangani vashanyi vakasiyana vanouya kwauri pamwedzi. Kuti ukwidze mutengo wekukwezva uye kuchengetedza, iwe unogona zvakare kuona shanduko yetraffic zvichienderana nezuva revhiki uye kunyangwe nguva yezuva.

MaFranchisor uye makambani emaketani anogona kuraira ongororo yakavakirwa pamifananidzo yemhando yemhando yezvitoro zvakasiyana siyana zvekutengesa: kuvapo kwemalogo, zviratidzo, maposta, mabhenji, zvichingodaro.

2.5. Nekutakura

Mumwe muenzaniso wekuona kuchengetedzwa uchishandisa mavhidhiyo analytics kuona zvinhu zvakasiiwa muhoro dzenhandare dzendege kana zviteshi zvechitima. Chiratidzo chinogona kudzidziswa kuziva zvinhu zvemazana emakirasi: zvidimbu zvefenicha, mabhegi, masutukesi, maamburera, marudzi akasiyana ezvipfeko, mabhodhoro, zvichingodaro. Kana vhidhiyo yako analytics system ikaona chinhu chisina muridzi uye ichiziva uchishandisa Vision, inotumira chiratidzo kune yekuchengetedza sevhisi. Basa rakafanana rinosanganiswa nekuona otomatiki mamiriro asina kujairika munzvimbo dzeveruzhinji: mumwe munhu anonzwa kurwara, kana mumwe munhu anoputa panzvimbo isiriyo, kana munhu anowira panjanji, uye zvichingodaro - ese aya mapatani anogona kuzivikanwa nevhidhiyo analytics masisitimu. kuburikidza neVision API.

2.6. Document flow

Chimwe chinonakidza chemangwana chekushandisa kweChiratidzo chatiri kugadzira parizvino kuzivikanwa kwegwaro uye kupatsanurwa kwavo otomatiki mumadhatabhesi. Panzvimbo pekupinda nemaoko (kana kuipa, kupinda) kusingaperi nhevedzano, manhamba, mazuva ekuburitswa, nhamba dzeakaundi, ruzivo rwebhangi, mazuva nenzvimbo dzekuzvarwa uye mamwe akawanda data akanyoreswa, unogona kuongorora magwaro uye wozvitumira otomatiki pamusoro pechiteshi chakachengeteka kuburikidza ne API kune gore, uko iyo sisitimu inozoziva magwaro aya panhunzi, patsanura uye wodzosera mhinduro ine data mune inodiwa fomati yekupinda otomatiki mudhatabhesi. Nhasi Vision inotoziva nzira yekuisa magwaro (kusanganisira PDF) - inosiyanisa pakati pepasipoti, SNILS, TIN, zvitupa zvekuzvarwa, zvitupa zvemuchato nezvimwe.

Ehe, iyo neural network haigone kubata ese aya mamiriro kunze kwebhokisi. Muchiitiko chega chega, modhi nyowani inovakwa kune chaiyo mutengi, zvinhu zvakawanda, nuances uye zvinodiwa zvinotariswa, seti yedata inosarudzwa, uye kudzokororwa kwekudzidziswa, kuyedzwa, uye kumisikidzwa kunoitwa.

3. API chirongwa chekushanda

Vision's "gedhi rekupinda" revashandisi ndiyo REST API. Inogona kugamuchira mafoto, vhidhiyo mafaera uye kutepfenyura kubva kunetiweki kamera (RTSP nzizi) sekuisa.

Kuti ushandise Vision, unoda kusaina vakakwira muMail.ru Cloud Solutions sevhisi uye gamuchira tokeni dzekuwana (mutengi_id + mutengi_secret). Kuvimbiswa kwemushandisi kunoitwa uchishandisa OAuth protocol. Iyo sosi data mumiviri yePOST zvikumbiro inotumirwa kuAPI. Uye mukupindura, mutengi anogashira kubva kuAPI mhedzisiro yekuzivikanwa muJSON fomati, uye mhinduro yacho yakarongeka: ine ruzivo nezve zvakawanikwa zvinhu uye marongero azvo.

Mhinduro yemuenzaniso

{
   "status":200,
   "body":{
      "objects":[
         {
            "status":0,
            "name":"file_0"
         },
         {
            "status":0,
            "name":"file_2",
            "persons":[
               {
                  "tag":"person9"
                  "coord":[149,60,234,181],
                  "confidence":0.9999,
                  "awesomeness":0.45
               },
               {
                  "tag":"person10"
                  "coord":[159,70,224,171],
                  "confidence":0.9998,
                  "awesomeness":0.32
               }
            ]
         }

         {
            "status":0,
            "name":"file_3",
            "persons":[
               {
               "tag":"person11",
               "coord":[157,60,232,111],
               "aliases":["person12", "person13"]
               "confidence":0.9998,
               "awesomeness":0.32
               }
            ]
         },
         {
            "status":0,
            "name":"file_4",
            "persons":[
               {
               "tag":"undefined"
               "coord":[147,50,222,121],
               "confidence":0.9997,
               "awesomeness":0.26
               }
            ]
         }
      ],
      "aliases_changed":false
   },
   "htmlencoded":false,
   "last_modified":0
}

Mhinduro ine inonakidza parameter kutyisa - ichi ndicho chinomisikidzwa che "kutonhorera" kwechiso mumufananidzo, nerubatsiro rwayo isu tinosarudza yakanakisa pfuti yechiso kubva mukutevedzana. Isu takadzidzisa neural network kufanotaura mukana wekuti foto ichafarirwa pasocial network. Izvo zvirinani kunaka kwemufananidzo uye kunyanya kunyemwerera kwechiso, kunowedzera kushamisa.

API Vision inoshandisa pfungwa inonzi space. Ichi chishandiso chekugadzira zviso zvakasiyana. Mienzaniso yenzvimbo ndeyemavara matema uye machena, mazita evashanyi, vashandi, vatengi, etc. Pachiratidzo chega chega muVision, unogona kugadzira nzvimbo dzinosvika gumi, nzvimbo yega yega inogona kusvika zviuru makumi mashanu zvePersonIDs, kureva, kusvika ku10 zviuru. per token . Uyezve, nhamba yematokeni paakaundi haina kuganhurirwa.

Nhasi API inotsigira nzira dzinotevera dzekuona uye dzekuzivikanwa:

• Ziva / Seta - kuona uye kuzivikanwa kwezviso. Otomatiki inopa PersonID kumunhu wese akasiyana, inodzosera iyo PersonID uye inoronga yevanowanikwa.
• Delete - kudzima yakananga PersonID kubva kudhatabhesi remunhu.
• Truncate - inobvisa nzvimbo yese kubva kuPersonID, inobatsira kana yakashandiswa senzvimbo yekuyedza uye unofanirwa kuseta zvakare dhatabhesi yekugadzira.
• Ziva - kucherechedzwa kwezvinhu, zviratidziro, marezinesi epureti, nzvimbo, mitsetse, nezvimwe. Inodzosa kirasi yezvinhu zvakawanikwa uye marongero azvo.
• Tsvaga zvinyorwa - inoona chaiwo marudzi ezvinyorwa zveRussian Federation (inosiyanisa pasipoti, SNILS, nhamba yekuzivikanwa yemutero, nezvimwewo).

Isu tichave zvakare munguva pfupi iri kupedzisa basa panzira dzeOCR, kuona kuti murume kana mukadzi, zera uye manzwiro, pamwe nekugadzirisa matambudziko ekutengesa, ndiko kuti, kudzora otomatiki kuratidzwa kwezvinhu muzvitoro. Iwe unogona kuwana yakazara API zvinyorwa pano: https://mcs.mail.ru/help/vision-api

4. Mhedziso

Ikozvino, kuburikidza neruzhinji API, unogona kuwana kuzivikanwa kwechiso mumifananidzo nemavhidhiyo; kuzivikanwa kwezvinhu zvakasiyana-siyana, marezenisi mahwendefa, marezinesi, zvinyorwa uye ese maficha anotsigirwa. Mamiriro ekushandisa - gungwa. Huya, uedze sevhisi yedu, isa iyo yakanyanya kuoma mabasa. Ekutanga 5000 kutengeserana ndeyemahara. Zvichida ichave "chipo chisipo" chezvirongwa zvako.

Unogona kuwana ipapo ipapo API pakunyoresa uye kubatana. kuona. Vese vashandisi veHabra vanogashira kodhi yekusimudzira yemamwe matransaction. Ndokumbira undinyorere email kero yawakashandisa kunyoresa account yako!

Source: www.habr.com