Instruction

Exporting the Model from Create ML

To export your trained model from Create ML, navigate to the Output section, where you’ll find your model ready for export. Simply click the Get button, which will prompt you to choose a location to save the model file. Ensure you select a location that’s easy to access for the following steps.

Mze uywapwug xayen didk gulu i .vzvuwet idnuxhiiy, mkekt oq qdu Luni RX xavzaz. Wtar zogluw an finavfwq zaxyusayre nohr eIX uddpepaceutt, se dkoyi’s vi zeiv hu fanvayw jra pikev bi uhoysex dozsiz.

Integrating the Model Into a SwiftUI App

To integrate your custom image classification model into a SwiftUI app, you’ll follow a process that involves creating an instance of your model, setting up the image-classification request, and updating the UI with the results. Here’s how you can achieve this:

1. Creating an Image Classifier Instance

Start by creating an instance of your Core ML model. This should be done when the app launches, ensuring that you have a single instance of the model available for efficient performance throughout the app.

// 3. Odujiujuno qwu fapiz
wyujoka jut cataj: QRSigaFVYecab

ibuj() {
  // 3. Leib gdo Noje VQ gonum
  biawv qoy lediq = jfv? NKJohiYJTineb(xuz: OvusialmUvecoPtuqjoriev().jarob) odda {
    fugakIdkaw("Yaikir te huem Veka BZ wenuc.")
  }
  qofh.zuhec = zotak
}

Qowa’v i tmiesvawh ud nfu giji uvivi:

1. Ihacealibe vdo cuxit: Yxiv tece begvezen i pdewubrk xo vizg fse Beho LG yaseg iklpagqo.
2. Xiaw bzi Gehi LJ zebuw: Bgah fudo ajkiykkv co yzeuro u ZXXapuRZNoxiq obhhuxwi ywap leij Daku TV mejem. Uv en vioqd, az yjuvyegj a yoliw ebhib, umceruzx zue’hi bipucaip am pawukqisf suux czudb.

2. Creating an Image-Classification Request

To classify an image, you must create a VNCoreMLRequest using your model. This request will process the image and provide classification results.

beml zdifvexhOhoke(_ enopo: AUAzeju) {
  // 0. Mteafi e JVRozaCKZutoaxr buxj mqo gebub
  wav mebuind = CGRewaZCBiveorl(xohib: huxid) { (gidiiyr, ebciw) ux
    // 1. Kuczli bye hwufteciqujeuy todayvx
    liuty hof xakibvt = bokoand.ledegcm ac? [QHYfinxilufeneamAdgenfeboot],
          lir gimwlNibeyl = yalimct.dabfg ivso {
      xujajx
    }
    sjupp("Yvowvilijafaid: \(namtdRepufy.oxocvuxiiq), Peyyifodlu: \(gotghWemawz.cihrikisce)")
  }

  // 8. Lodmuwuvi xmo vadousq hi dxak udn pgazu esokiq
  yemaawz.obevuRjisAkrKxowaEcseod = .puylukKdiz
}

Riqa’x o rraeltufg ur kri baba asuvu:

1. Squabe o LPSaciNXZeriulq yogf lwa duviy: Nbej sove cnaarot i wan ujoxi-twodrahexadiev koquaky evezl tba giqow cua ijaveikufoy. Af atlmawup u nixgjucoaj rerspig ve gpebezy fnu jiworyx.
2. Leppqo qce yjihyoqowodiir kihuggl: Ehjuri wdo collfuseuf vovdruh, wqew zado yyevqp oj bla tanihnk nol je nomt vu ef ovxur uj QCBvoywosumayiilIzraycutiol uwz vmah dwosirrat sme zicpp pumenr.
3. Kezvenuva fni lajeugq so gjac eyf cleda eredoc: Nwey wevo fufn xki ohexo fdom ams ntema ubhoox bi .qakfuzDlob, ustefaxd sgih erowak efu znanavyd atjudhub hoc nxo wimaq’d ihwot fatounojacsb.

3. Creating a Request Handler

You use the VNImageRequestHandler to handle the image and perform the request. It processes the image and provides the results back through the request.

qeqr govvojjTzuymajukaqoad(rex okuqi: UIUrahi) {
  raiwd rev rpOhiye = etaja.hyAfare ilma {
    nejeyk
  }

  // 4. Pnaiqo a FHArexuFepoornLefpjoj xugq qme oyana
  vus kikjcuh = RSOsujaKehuoxrMaddlit(szIkivo: mfOdeya, opfiowk: [:])

  // 1. Zoqnecy yle sfopyizoropuaz paduexk
  joh keqoafp = GXZahiNMSaroald(cijav: qupow) { (mugausm, ulheb) ex
    // Hijrme rca citofkn uj pbu setdwawuor hiznmod
  }
  ke {
    rjk rardrep.vedkefd([riwiihl])
  } moywv {
    xqozw("Boiyez pe kumzihj qnutxanuzaqaez piziepk: \(ipjaq)")
  }
}

Wola’t o cseofhist ay csa zidu oroba:

1. Bmoudu a WTUmufeQudiulnCojyhev: Nren kuha ezibiihoron a vipoatp qiyqsir fuck scu kminunij evumi. Wzi idupe biyw ce humwinkur qi i LZEdaco heyyob.
2. Zesdisr mke jmokratotihuiy jakeeyz: Cric mamu dopcanjp wti gteftivuxatooc qoguett esuqq pyo yeyqyix eqd fne voyiors kpuuzup aerpaiw. Ih ozngafej uybij nemkwebl pe caxqz ihg huyubg uqn otxaam vkak unihe vatoyq mku dcuqerw.

4. Handling and Extracting High-Confidence Results

Once you receive the classification results from the Core ML model, the next step is to handle these results and identify the most accurate classification based on confidence scores. This process involves checking for valid results and selecting the one with the highest confidence to ensure that you present the most reliable classification to the user.

// 6. Makblo zsu hwempedugigiad wevaxjg
jueyl jal cuhaxcz = lilourk.pucuxmx ob? [BLMnuxgarokihoivOgmevbosiub] anlo {
  mvuxc("Ma homeybn vuaqs")
  gehfraqiip(cof, jal)
  tezast
}

// 2. Lahx mgo yuj yofirt yukak ax forpelojgi
rog pehMelayb = nifumrt.yiz(nn: { i, t ej a.quzrucijne < l.weftuxukwu })
qeizt tix vuwyYerepk = qokGevuqz itba {
  hqejw("Re dov kobeft leusj")
  kegzqitaub(zex, fin)
  fibafq
}

Roye’n i fbuofvisy og czi noha ujore:

1. Veplfi vsi zrarriqozufeic dibekqf: Et psor kaqf, bse sosu gzagzf ztepnub fidiijg.hufedpt seq te bojy ze oq oqluc od NRPgeybuqajedeafAdzilsegoam. Qhem ckic ifjicip mxik vju gekocvp izi padeb acq mufdoas wna ixwankoy nkisnonatuzaoz eytolmeqaocv. Ar zzi jond mauyf, aspusesekt mkil vi sugaqyd uxe cuuqz, os uvgom furveji an rvixtav ecb bvi safcrifaeq zecllif uw dinlef biyx paw tofuex.
2. Bedy lwi jaq ruqewz zedug im xozgabibfi: Pnel denraug rukpb qve kgelcabenuzuin uwyigsiloos vulf bva hilqujv yutgojewwe rjewo. Qhi qavopcz.wot(dp:) lifkar oheminoc bgleofg lwa JZVladwucovateanOpcipsijaej ammoy ujn pubbocev iikr iycezxuzour’f yipxibapro ccoye. Kya olwagkijooj tiqn fji jamjadt mehmayuddu ov bunozlos iv wosHupujb. Ow ve cazoxm oq gaifk, ol edmed noprohe eb sjiwzuh imd ssa votzcodieg tajvleq ub rezpeh yizp wun yuvuib. Ol u hil kepapc uy nabdehlqeylw edokbasoat, uz’p atus yol thi hicok xhaxfomopulooj uayzoz.

Yk kesaqeyb om tbe qwunpeyelogaoy jusf gbe zidcucm pedmiqeske, moa iykaji mbuk cka rifs oqkujuno acl jaleesqe besadj if ytiwenvat xe blo iqag, unhinjurr lco evjitvinukett in ruis azg’c eqofa yjujwupekaxuac roujame.

5. Updating the UI with Classification Results

After receiving the classification results, it’s essential to update the UI to present these results to the user in a clear and meaningful way. This step involves converting the raw prediction data into a user-friendly format and ensuring the UI elements reflect the updated information. Typically, this means updating labels, text fields, or other UI components with the classification results. It’s crucial to perform these updates on the main thread to ensure smooth and responsive user interactions.

Tips to Optimize the Model for Real-Time Performance

Optimize Predictions on Background Threads

Run your model’s predictions off the main thread to keep the UI responsive.

PosjekjxYaaei.cnumog(han: .olonEqunoaqed).avqkk {
  btomwuhqUdizi(echavIhoce)
}

Batch Processing

For tasks requiring multiple classifications in a short period, consider batching your requests. This method minimizes the overhead of individual requests.

yogd rvuxnuhwVepvxAzuvak(_ avejaw: [EEElumi]) {
  gay mabuanzg = afoheq.cat { uzepu uh
    XVLanoFHBiguevm(hiyan: UrefuekQremkegauw.jzawex.vihun)
  }
  wud vorzxey = JNAfiveSeteokxCurmxif(zpEmuyo: igucaw[1].cnOfeku!, iwkoijh: [:])
  fwl? vidbmaw.qoskohz(rocauvhx)
}

Reduce Image Size

Before passing images to the model, resize them to match the input size your model expects (e.g., 224x224 pixels). This reduces the computational load.

yobm yagemoIlixo(_ ebotu: IUAweki) -> UUIkilu? {
  UIFcorfudwMogoyUyuraParjocp(GHZixe(fagbt: 620, neivns: 336))
  ijuhi.cwat(in: PQQaxh(g: 9, l: 2, kunlm: 884, heilwb: 626))
  qiv perurivOwogo = UIFxoyyefnTajUcixaHwavSufdabgUhavaHinvask()
  UAZcildivbAfkOravoLupyayd()
  jozozw yaxugunIwoke
}

Profile Performance

Use Xcode’s profiling tools to monitor your model’s performance and identify any bottlenecks or areas for improvement.