Cov thev naus laus zis thiab cov qauv rau peb lub khoos phis tawj yav tom ntej tau tsim thiab txhim kho maj mam thiab hauv cov haujlwm sib txawv ntawm peb lub tuam txhab - hauv Mail, Huab, Tshawb Nrhiav. Lawv matured zoo li cheese los yog cognac. Muaj ib hnub peb pom tau hais tias peb cov neural networks qhia tau zoo heev hauv kev lees paub, thiab peb tau txiav txim siab los muab lawv rau hauv ib qho khoom b2b - Lub Zeem Muag - uas tam sim no peb siv peb tus kheej thiab muab rau koj siv.
Niaj hnub no, peb lub computer tsis pom kev thev naus laus zis ntawm Mail.Ru Cloud Solutions platform tau ua tiav kev ua haujlwm thiab daws cov teeb meem nyuaj heev. Nws yog raws li ib tug xov tooj ntawm neural networks uas tau kawm nyob rau hauv peb cov ntaub ntawv teev thiab tshwj xeeb hauv kev daws teeb meem siv. Tag nrho cov kev pab cuam khiav ntawm peb cov chaw server. Koj tuaj yeem koom ua ke rau pej xeem Vision API rau hauv koj daim ntawv thov, dhau los ntawm txhua lub peev xwm ntawm cov kev pabcuam muaj. API yog nrawm - ua tsaug rau cov neeg rau zaub mov GPUs, lub sijhawm teb nruab nrab hauv peb lub network yog 100 ms.
Mus rau miv, muaj cov ncauj lus kom ntxaws thiab ntau yam piv txwv ntawm Lub Zeem Muag txoj haujlwm.
Ib qho piv txwv ntawm qhov kev pabcuam uas peb tus kheej siv cov cuab yeej hais txog lub ntsej muag lees paub yog . Ib qho ntawm nws cov khoom yog Vision photo stands, uas peb nruab ntawm ntau lub rooj sib tham. Yog tias koj mus txog qhov chaw yees duab, thaij duab nrog lub koob yees duab built-in thiab nkag mus rau koj tus email, lub kaw lus yuav pom tam sim ntawm cov duab ntawm cov duab uas koj tau ntes los ntawm cov neeg ua haujlwm ntawm lub rooj sib tham, thiab, yog tias xav tau, yuav xa cov duab pom rau koj los ntawm email. Thiab peb tsis tau tham txog kev txhaj tshuaj portrait - Lub zeem muag pom koj txawm nyob hauv keeb kwm yav dhau los ntawm cov neeg tuaj saib. Tau kawg, nws tsis yog daim duab sawv cev rau lawv tus kheej uas tau lees paub, cov no tsuas yog cov ntsiav tshuaj hauv cov chaw zoo nkauj uas tsuas yog thaij duab ntawm cov qhua nrog lawv lub koob yees duab built-in thiab xa cov ntaub ntawv mus rau cov servers, qhov twg txhua qhov kev paub txog khawv koob tshwm sim. Thiab peb tau pom ntau dua li qhov xav tsis thoob ntawm kev siv thev naus laus zis ua tau zoo txawm tias ntawm cov kws paub txog duab tshwj xeeb. Hauv qab no peb yuav tham txog qee qhov piv txwv.
1. Peb lub ntsej muag paub tus qauv
1.1. Neural network thiab kev ua haujlwm ceev
Rau kev lees paub, peb siv qhov kev hloov kho ntawm ResNet 101 neural network qauv. Qhov nruab nrab Pooling qhov kawg yog hloov los ntawm cov txheej txheem sib txuas, zoo ib yam li yuav ua li cas hauv ArcFace. Txawm li cas los xij, qhov loj ntawm cov duab sawv cev yog 128, tsis yog 512. Peb qhov kev cob qhia muaj txog 10 lab daim duab ntawm 273 tus neeg.
Tus qauv khiav ceev heev ua tsaug rau kev ua tib zoo xaiv server configuration architecture thiab GPU xam. Nws yuav siv sij hawm los ntawm 100 ms kom tau txais cov lus teb los ntawm API ntawm peb cov tes hauj lwm sab hauv - qhov no suav nrog kev kuaj pom lub ntsej muag (kuaj lub ntsej muag hauv daim duab), lees paub thiab xa rov qab PersonID hauv API teb. Nrog loj ntim ntawm cov ntaub ntawv tuaj - duab thiab yeeb yaj duab - nws yuav siv sij hawm ntau npaum li cas los hloov cov ntaub ntawv mus rau qhov kev pab cuam thiab kom tau txais cov lus teb.
1.2. Kev ntsuas qhov ua tau zoo ntawm tus qauv
Tab sis kev txiav txim siab qhov ua tau zoo ntawm neural networks yog ib txoj haujlwm tsis meej heev. Qhov zoo ntawm lawv txoj haujlwm yog nyob ntawm seb cov ntaub ntawv teev cov qauv tau kawm thiab seb lawv puas tau ua kom zoo rau kev ua haujlwm nrog cov ntaub ntawv tshwj xeeb.
Peb pib ntsuas qhov tseeb ntawm peb cov qauv nrog cov kev xeem LFW nrov, tab sis nws yog me me thiab yooj yim heev. Tom qab ncav cuag 99,8% qhov tseeb, nws tsis muaj txiaj ntsig ntxiv lawm. Muaj kev sib tw zoo rau kev ntsuas cov qauv kev lees paub - Megaface, uas peb maj mam mus txog 82% qeb 1. Kev xeem Megaface muaj ib lab daim duab - cuam tshuam - thiab tus qauv yuav tsum muaj peev xwm paub qhov txawv ntau txhiab daim duab ntawm cov neeg muaj npe nrov ntawm Facescrub. dataset los ntawm distractors. Txawm li cas los xij, tau tshem tawm Megaface qhov kev xeem ntawm qhov tsis raug, peb pom tias nrog cov ntawv tshem tawm peb ua tiav qhov tseeb ntawm 98% qib 1 (cov duab ntawm cov neeg muaj koob muaj npe feem ntau yog qhov tshwj xeeb). Yog li ntawd, lawv tau tsim ib qho kev xeem sib cais, zoo ib yam li Megaface, tab sis nrog cov duab ntawm cov neeg "zoo tib yam". Tom qab ntawd peb tau txhim kho qhov kev lees paub qhov tseeb ntawm peb cov ntaub ntawv thiab mus deb dhau mus. Tsis tas li ntawd, peb siv qhov kev ntsuam xyuas zoo uas muaj ntau txhiab daim duab; nws simulates ntsej muag tagging hauv cov neeg siv huab. Hauv qhov no, pawg yog pawg ntawm cov tib neeg zoo sib xws, ib pawg rau txhua tus neeg paub. Peb tau txheeb xyuas qhov ua tau zoo ntawm kev ua haujlwm ntawm pab pawg tiag tiag (tseeb).
Tau kawg, kev paub yuam kev tshwm sim nrog txhua tus qauv. Tab sis cov xwm txheej zoo li no feem ntau raug daws los ntawm kev nplua qhov pib rau cov xwm txheej tshwj xeeb (rau txhua lub rooj sib tham peb siv tib qhov chaw pib, tab sis, piv txwv li, rau kev tswj hwm kev tswj hwm peb yuav tsum tau ua kom muaj kev nce qib kom muaj tsawg dua qhov tsis zoo). Feem coob ntawm cov neeg tuaj xyuas lub rooj sib tham tau lees paub raug los ntawm peb lub Vision photo booths. Qee lub sij hawm ib tug neeg yuav saib cov qoob loo saib ua ntej thiab hais tias, "Koj lub kaw lus ua yuam kev, nws tsis yog kuv." Tom qab ntawd peb qhib daim duab tag nrho, thiab nws tau muab tawm tias muaj tiag tiag yog tus qhua hauv daim duab no, tsuas yog peb tsis tau yees duab nws, tab sis lwm tus, tus neeg tsuas yog tshwm sim hauv keeb kwm yav dhau ntawm thaj chaw qhov muag plooj. Ntxiv mus, lub neural network feem ntau raug lees paub txawm tias ib feem ntawm lub ntsej muag tsis pom, lossis tus neeg sawv hauv qhov profile, lossis txawm tias ib nrab tig. Lub kaw lus tuaj yeem paub txog ib tus neeg txawm tias lub ntsej muag nyob hauv thaj tsam ntawm qhov muag tsis pom kev, hais tias, thaum tua nrog lub lens dav.
1.3. Piv txwv ntawm kev sim hauv cov xwm txheej nyuaj
Hauv qab no yog cov piv txwv ntawm peb cov neural network ua haujlwm li cas. Cov duab raug xa mus rau cov tswv yim, uas nws yuav tsum sau npe siv PersonID - tus cim tshwj xeeb ntawm tus neeg. Yog tias ob lossis ntau cov duab muaj tib tus ID, ces, raws li cov qauv, cov duab no qhia txog tib neeg.
Cia peb tam sim ntawd nco ntsoov tias thaum kuaj, peb muaj kev nkag mus rau ntau yam tsis muaj thiab cov qauv pib uas peb tuaj yeem teeb tsa kom ua tiav ib qho txiaj ntsig. Cov pej xeem API tau ua kom zoo rau qhov raug siab tshaj plaws ntawm cov xwm txheej.
Cia peb pib nrog qhov yooj yim tshaj plaws, nrog kev paub txog lub ntsej muag pem hauv ntej.
Zoo, qhov ntawd yooj yim heev. Cia peb ua haujlwm nyuaj, ntxiv cov hwj txwv thiab ib qho puv xyoo.
Ib txhia yuav hais tias qhov no kuj tsis nyuaj, vim hais tias nyob rau hauv ob qho tib si lub ntsej muag tag nrho yog pom, thiab ntau cov ntaub ntawv hais txog lub ntsej muag muaj nyob rau hauv lub algorithm. Okay, cia peb tig Tom Hardy rau hauv profile. Qhov teeb meem no yog qhov nyuaj dua, thiab peb siv ntau lub dag zog los daws nws thaum tswj qhov kev ua yuam kev tsawg: peb xaiv cov txheej txheem kev cob qhia, xav los ntawm kev tsim qauv ntawm cov neural network, honed lub zog poob thiab txhim kho qhov ua ntej. ntawm cov duab.
Cia peb muab ib lub headdress rau nws:
Los ntawm txoj kev, qhov no yog ib qho piv txwv ntawm ib qho teeb meem tshwj xeeb, vim lub ntsej muag yog hnyav heev, thiab nyob rau hauv daim duab hauv qab kuj tseem muaj duab ntxoov ntxoo sib sib zog nqus zais qhov muag. Nyob rau hauv lub neej tiag tiag, tib neeg feem ntau hloov lawv cov tsos nrog kev pab los ntawm cov tsom iav dub. Cia peb ua tib yam nrog Tom.
Okay, cia peb sim pov cov duab los ntawm cov hnub nyoog sib txawv, thiab lub sijhawm no peb yuav sim nrog lwm tus neeg ua yeeb yam. Cia peb ua piv txwv ntau dua, qhov uas muaj hnub nyoog txog kev hloov pauv tshwj xeeb tshaj tawm. Qhov xwm txheej tsis yog nyob deb, nws tshwm sim ntau zaus thaum koj xav tau los sib piv cov duab hauv daim ntawv hla tebchaws nrog lub ntsej muag ntawm tus tuav. Tom qab tag nrho, thawj daim duab yog ntxiv rau daim ntawv hla tebchaws thaum tus tswv hnub nyoog 20 xyoo, thiab thaum muaj hnub nyoog 45 xyoo ib tus neeg tuaj yeem hloov pauv tau zoo:
Koj puas xav tias tus kws tshaj lij tseem ceeb ntawm kev tshaj tawm txoj moo zoo tsis tau hloov pauv ntau nrog hnub nyoog? Kuv xav tias txawm tias ob peb tus neeg yuav muab cov duab saum toj thiab hauv qab, tus tub hluas tau hloov pauv ntau xyoo.
Neural tes hauj lwm ntsib kev hloov pauv ntawm cov tsos mob ntau dua. Piv txwv li, qee zaum cov poj niam tuaj yeem hloov lawv cov duab zoo heev nrog kev pab ntawm cov tshuaj pleev ib ce:
Tam sim no cia peb ua haujlwm nyuaj ntau dua: xav tias qhov sib txawv ntawm lub ntsej muag tau npog hauv cov duab sib txawv. Hauv qhov no, algorithm tsis tuaj yeem sib piv tag nrho cov qauv. Txawm li cas los xij, Vision tuav cov xwm txheej zoo li no.
Los ntawm txoj kev, tuaj yeem muaj ntau lub ntsej muag hauv daim duab; piv txwv li, ntau tshaj 100 tus neeg tuaj yeem haum rau hauv daim duab dav dav ntawm ib lub tsev. Qhov no yog qhov teeb meem nyuaj rau neural tes hauj lwm, txij li ntau lub ntsej muag tuaj yeem teeb qhov sib txawv, qee qhov tsis pom tseeb. Txawm li cas los xij, yog tias daim duab tau txais kev daws teeb meem txaus thiab zoo (tsawg kawg 75 pixels ib square npog lub ntsej muag), Lub Zeem Muag yuav tuaj yeem ntes thiab paub txog nws.
Qhov peculiarity ntawm kev tshaj tawm cov duab thiab cov duab los ntawm cov koob yees duab soj ntsuam yog tias tib neeg feem ntau plooj vim lawv tsis pom tseeb lossis txav mus rau lub sijhawm ntawd:
Tsis tas li ntawd, lub teeb pom kev zoo tuaj yeem sib txawv ntawm cov duab mus rau duab. Qhov no, ib yam nkaus thiab, feem ntau dhau los ua qhov tsis txaus ntseeg; ntau lub algorithms muaj qhov nyuaj heev kom ua tiav cov duab uas tsaus dhau thiab lub teeb dhau, tsis txhob hais kom raug txuam nrog lawv. Cia kuv ceeb toom rau koj tias kom ua tiav qhov txiaj ntsig no koj yuav tsum teeb tsa qhov pib hauv ib txoj hauv kev; cov yam ntxwv no tseem tsis tau muaj rau pej xeem. Peb siv tib lub neural network rau txhua tus neeg siv khoom; nws muaj qhov pib uas tsim nyog rau feem ntau cov dej num.
Peb nyuam qhuav dov tawm ib qho tshiab ntawm tus qauv uas lees paub Asian ntsej muag nrog cov neeg siab. Qhov no siv los ua ib qho teeb meem loj, uas tseem hu ua "kev kawm tshuab" (lossis "neural network") kev ntxub ntxaug. European thiab Asmeskas neural tes hauj lwm tau lees paub Caucasian ntsej muag zoo, tab sis nrog Mongoloid thiab Negroid ntsib qhov xwm txheej phem dua. Tej zaum, nyob rau hauv Suav teb qhov teeb meem yog raws nraim qhov opposite. Nws yog txhua yam hais txog kev cob qhia cov ntaub ntawv uas cuam tshuam txog cov hom neeg tseem ceeb hauv ib lub tebchaws. Txawm li cas los xij, qhov xwm txheej hloov pauv; niaj hnub no qhov teeb meem no tsis mob heev. Lub zeem muag tsis muaj teeb meem nrog cov neeg sib txawv.
Kev lees paub lub ntsej muag tsuas yog ib qho ntawm ntau daim ntawv thov ntawm peb cov thev naus laus zis; Lub zeem muag tuaj yeem raug cob qhia kom paub txog txhua yam. Piv txwv li, daim ntawv tso cai, suav nrog hauv cov xwm txheej nyuaj rau algorithms: ntawm cov ces kaum ntse, qias neeg thiab nyuaj nyeem daim ntawv tso cai.
2. Cov ntaub ntawv siv tau
2.1. Kev tswj kev siv lub cev: thaum ob tus neeg siv tib daim ntawv hla tebchaws
Nrog kev pab los ntawm Lub Zeem Muag, koj tuaj yeem siv cov txheej txheem rau kev sau cov neeg ua haujlwm tuaj txog thiab tawm mus. Cov txheej txheem ib txwm ua raws li kev hla hluav taws xob muaj qhov tsis zoo, piv txwv li, koj tuaj yeem hla ob tus neeg siv ib daim paib. Yog tias qhov system tswj kev nkag mus (ACS) tau ntxiv nrog Lub Zeem Muag, nws yuav ncaj ncees sau leej twg tuaj / sab laug thiab thaum twg.
2.2. Sijhawm taug qab
Qhov kev siv lub zeem muag no muaj feem cuam tshuam nrog rau yav dhau los. Yog tias koj ntxiv cov txheej txheem nkag nrog peb qhov kev pabcuam lub ntsej muag, nws yuav tuaj yeem tsis tsuas yog txhawm rau txhawm rau txhawm rau txhawm rau txhawm rau txhawm rau txhawm rau txhawm rau txhawm rau txhawm rau txhawm rau txhawm rau txhawm rau txhawm rau txhawm rau txhawm rau txhawm rau txhawm rau txhawm rau txhawm rau txhawm rau txheeb xyuas qhov tseeb ntawm cov neeg ua haujlwm hauv lub tsev lossis chaw. Hauv lwm lo lus, Lub Zeem Muag yuav pab koj ua siab ncaj rau hauv tus account tias leej twg tuaj ua haujlwm thiab tawm thaum lub sijhawm twg, thiab leej twg hla kev ua haujlwm tag nrho, txawm tias nws cov npoj yaig tau them rau nws rau pem hauv ntej ntawm nws cov thawj coj.
2.3. Video Analytics: Tib neeg taug qab thiab kev nyab xeeb
Los ntawm kev taug qab cov neeg siv lub zeem muag, koj tuaj yeem ntsuas qhov tseeb ntawm cov tsheb thauj mus los ntawm cov chaw yuav khoom, chaw nres tsheb ciav hlau, cov kab, txoj kev thiab ntau qhov chaw pej xeem. Peb qhov kev taug qab kuj tseem tuaj yeem pab tau zoo hauv kev tswj kev nkag mus, piv txwv li, mus rau lub tsev khaws khoom lossis lwm qhov chaw ua haujlwm tseem ceeb. Thiab tau kawg, taug qab tib neeg thiab ntsej muag pab daws teeb meem kev nyab xeeb. Tau ntes ib tug neeg nyiag ntawm koj lub khw? Ntxiv nws tus PersonID, uas tau rov qab los ntawm Lub Zeem Muag, rau hauv daim ntawv teev npe dub ntawm koj qhov kev tshuaj ntsuam video software, thiab lwm zaus lub kaw lus yuav ceeb toom kev nyab xeeb tam sim yog tias hom no tshwm sim dua.
2.4. Hauv kev lag luam
Cov khw muag khoom thiab ntau yam kev pabcuam kev lag luam yog txaus siab rau kev lees paub txog kab. Nrog kev pab los ntawm Lub Zeem Muag, koj tuaj yeem lees paub tias qhov no tsis yog cov neeg coob coob, tab sis ib kab, thiab txiav txim siab nws qhov ntev. Thiab tom qab ntawd lub kaw lus ceeb toom rau cov neeg saib xyuas txog ib kab kom lawv tuaj yeem txheeb xyuas qhov xwm txheej: txawm tias muaj cov neeg tuaj xyuas thiab cov neeg ua haujlwm ntxiv yuav tsum raug hu, lossis ib tus neeg poob rau lawv txoj haujlwm.
Lwm txoj haujlwm nthuav yog cais cov neeg ua haujlwm hauv tuam txhab hauv chav los ntawm cov qhua. Feem ntau, lub kaw lus tau raug cob qhia kom cais cov khoom hnav khaub ncaws (txoj cai hnav khaub ncaws) lossis nrog qee qhov tshwj xeeb (cov phuam qhwv caj dab, ntawv cim rau hauv siab, thiab lwm yam). Qhov no pab txhawm rau txheeb xyuas cov neeg tuaj koom kom raug ntau dua (kom cov neeg ua haujlwm tsis "ua kom" cov ntaub ntawv ntawm cov neeg hauv tsev los ntawm lawv qhov chaw nyob).
Siv lub ntsej muag lees paub, koj tuaj yeem ntsuas koj cov neeg tuaj saib: qhov kev ntseeg siab ntawm cov neeg tuaj saib yog dab tsi, uas yog, muaj pes tsawg tus neeg rov qab los rau koj lub lag luam thiab nrog qhov zaus. Xam pes tsawg tus qhua tshwj xeeb tuaj rau koj hauv ib hlis. Txhawm rau txhim kho tus nqi ntawm kev nyiam thiab khaws cia, koj tuaj yeem pom qhov kev hloov pauv ntawm kev khiav tsheb nyob ntawm hnub ntawm lub lim tiam thiab txawm tias lub sijhawm nruab hnub.
Cov tuam txhab lag luam thiab cov tuam txhab lag luam tuaj yeem xaj qhov kev ntsuam xyuas raws li cov duab ntawm cov khoom lag luam zoo ntawm ntau lub khw muag khoom: muaj lub logo, cov paib, cov ntawv tshaj tawm, chij, thiab lwm yam.
2.5. Los ntawm kev thauj mus los
Lwm qhov piv txwv ntawm kev ua kom muaj kev ruaj ntseg siv kev tshuaj ntsuam video yog txheeb xyuas cov khoom pov tseg hauv cov tsev hauv tshav dav hlau lossis chaw tsheb ciav hlau. Lub zeem muag tuaj yeem raug cob qhia kom paub txog cov khoom ntawm ntau pua chav kawm: tej rooj tog, hnab, hnab, hnab, kaus, ntau hom khaub ncaws, lub raj mis, thiab lwm yam. Yog tias koj qhov kev tshuaj ntsuam video pom ib yam khoom tsis muaj tswv thiab pom nws siv Vision, nws xa cov teeb liab mus rau qhov kev pabcuam kev nyab xeeb. Ib txoj haujlwm zoo sib xws yog cuam tshuam nrog kev tsis siv neeg tshawb nrhiav qhov xwm txheej tsis zoo hauv cov chaw pej xeem: ib tus neeg hnov ββββmob, lossis ib tus neeg haus luam yeeb hauv qhov tsis ncaj ncees lawm, lossis ib tus neeg ntog rau ntawm txoj hlua, thiab lwm yam - tag nrho cov qauv no tuaj yeem lees paub los ntawm kev tshuaj ntsuam video. los ntawm Vision API.
2.6. Cov ntaub ntawv ntws
Lwm qhov kev nthuav dav yav tom ntej ntawm Lub Zeem Muag uas peb tab tom tsim tam sim no yog kev lees paub cov ntaub ntawv thiab lawv tsis siv neeg parsing rau hauv databases. Es tsis txhob manually nkag mus (lossis phem dua, nkag mus) tsis kawg series, cov lej, hnub teeb meem, tus lej nyiaj, cov ntsiab lus hauv tuam txhab nyiaj, hnub yug thiab chaw yug thiab ntau lwm cov ntaub ntawv raug cai, koj tuaj yeem luam theej duab cov ntaub ntawv thiab xa lawv mus rau qhov chaw ruaj ntseg ntawm lub API rau huab, qhov twg lub kaw lus yuav lees paub cov ntaub ntawv no ntawm ya, txheeb xyuas lawv thiab xa cov lus teb nrog cov ntaub ntawv hauv cov ntawv xav tau rau kev nkag mus rau hauv cov ntaub ntawv tsis siv neeg. Niaj hnub no Vision twb paub yuav ua li cas cais cov ntaub ntawv (xws li PDF) - qhov txawv ntawm daim ntawv hla tebchaws, SNILS, TIN, daim ntawv yug me nyuam, daim ntawv pov thawj kev sib yuav thiab lwm yam.
Ntawm chav kawm, lub neural network tsis muaj peev xwm daws tau tag nrho cov xwm txheej no tawm ntawm lub thawv. Nyob rau hauv txhua rooj plaub, tus qauv tshiab yog tsim rau cov neeg siv khoom tshwj xeeb, ntau yam, nuances thiab cov kev xav tau raug coj mus rau hauv tus account, cov ntaub ntawv raug xaiv, thiab rov ua dua ntawm kev cob qhia, kuaj, thiab teeb tsa.
3. API txoj haujlwm ua haujlwm
Lub zeem muag "qhov rooj nkag" rau cov neeg siv yog REST API. Nws tuaj yeem tau txais cov duab, cov ntaub ntawv video thiab tshaj tawm los ntawm lub koob yees duab network (RTSP kwj) raws li cov tswv yim.
Txhawm rau siv Vision, koj xav tau hauv Mail.ru Cloud Solutions cov kev pabcuam thiab tau txais tokens (client_id + client_secret). Cov neeg siv authentication yog ua los ntawm OAuth raws tu qauv. Cov ntaub ntawv hauv lub cev ntawm POST thov raug xa mus rau API. Thiab hauv cov lus teb, tus neeg siv khoom tau txais los ntawm API qhov kev lees paub hauv JSON hom, thiab cov lus teb tau tsim kho: nws muaj cov ntaub ntawv hais txog cov khoom pom thiab lawv cov kev tswj hwm.
Qauv teb
{
"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
}Cov lus teb muaj qhov nthuav tsis txaus ntseeg - qhov no yog qhov xwm txheej "txias" ntawm lub ntsej muag hauv daim duab, nrog rau nws pab peb xaiv qhov zoo tshaj plaws txhaj tshuaj ntawm lub ntsej muag los ntawm qhov sib lawv liag. Peb tau cob qhia neural network los twv seb qhov yuav ua li cas thiaj li yuav nyiam cov duab hauv social network. Qhov zoo dua qhov zoo ntawm daim duab thiab ntau lub ntsej muag luag ntxhi, qhov zoo dua qhov txaus ntshai.
API Vision siv lub tswvyim hu ua qhov chaw. Qhov no yog ib qho cuab yeej tsim kom muaj ntau lub ntsej muag. Piv txwv ntawm qhov chaw yog cov npe dub thiab dawb, cov npe ntawm cov qhua, cov neeg ua haujlwm, cov neeg siv khoom, thiab lwm yam. Rau txhua qhov token hauv Vision, koj tuaj yeem tsim txog li 10 qhov chaw, txhua qhov chaw tuaj yeem muaj txog 50 txhiab PersonIDs, uas yog, txog li 500 txhiab. rau token. Ntxiv mus, tus naj npawb ntawm tokens ib tus account tsis txwv.
Niaj hnub no API txhawb cov kev tshawb nrhiav thiab kev lees paub hauv qab no:
- Paub / Teeb - nrhiav kom pom thiab paub txog lub ntsej muag. Yuav muab tus PersonID rau txhua tus neeg tshwj xeeb, xa rov qab PersonID thiab kev tswj hwm ntawm cov neeg pom.
- Rho tawm - rho tawm ib tus neeg ID tshwj xeeb los ntawm tus neeg database.
- Truncate - tshem tag nrho qhov chaw los ntawm PersonID, pab tau yog tias nws tau siv los ua qhov chaw sim thiab koj yuav tsum rov pib dua cov ntaub ntawv rau kev tsim khoom.
- Detect - nrhiav pom cov khoom, scenes, daim ntawv tso cai, landmarks, queues, thiab lwm yam
- Ntes rau cov ntaub ntawv - kuaj pom cov ntaub ntawv tshwj xeeb ntawm Lavxias Federation (qhov txawv ntawm daim ntawv hla tebchaws, SNILS, tus lej cim se, thiab lwm yam).
Peb kuj tseem yuav ua tiav sai sai ntawm txoj hauv kev rau OCR, txiav txim siab poj niam txiv neej, hnub nyoog thiab kev xav, nrog rau kev daws teeb meem kev muag khoom, uas yog, kom tswj tau cov khoom muag hauv khw muag khoom. Koj tuaj yeem pom cov ntaub ntawv API tiav ntawm no:
4. Xaus
Tam sim no, los ntawm pej xeem API, koj tuaj yeem nkag mus rau lub ntsej muag lees paub hauv cov duab thiab yeeb yaj kiab; kev txheeb xyuas ntau yam khoom, daim ntawv tso cai, thaj chaw, cov ntaub ntawv thiab tag nrho cov xwm txheej tau txais kev txhawb nqa. Daim ntawv thov scenarios - hiav txwv. Tuaj, sim peb cov kev pabcuam, teeb tsa nws cov haujlwm nyuaj tshaj plaws. Thawj 5000 kev hloov pauv yog pub dawb. Tej zaum nws yuav yog "cov khoom tsis muaj" rau koj cov haujlwm.
Koj tuaj yeem nkag mus rau API tam sim ntawd thaum sau npe thiab kev sib txuas. . Txhua tus neeg siv Habra tau txais tus lej tshaj tawm rau kev hloov pauv ntxiv. Thov sau rau kuv tus email chaw nyob uas koj tau siv los sau npe koj tus account!
Tau qhov twg los: www.hab.com