Instruction

Now that you have a model, you can integrate this model into your iOS app. Find and open the Starter project for this lesson. You can find the materials for this project in the Walkthrough folder. Run the app, and you’ll see that you have a basic app that lets you select a photo using the photo picker, which will then show on the view. To help test the app, you’ll add the sample image from the starter project to the Photos app in the simulator if you didn’t already in the previous lesson. Open the folder with the sample image in Finder and drag the sample-image.jpg file onto the iOS simulator.

Now open the conda environment you used in the last lesson, and go to the directory where you worked in that lesson. Start the Python interpreter and enter the following one line at a time:

from ultralytics import YOLO
import os

model = YOLO("yolov8x-oiv7")
model.export(format="coreml", nms=True, int8=True)
os.rename("yolov8x-oiv7.mlpackage", "yolov8x-oiv7-int.mlpackage")
model.export(format="coreml", nms=True)

model = YOLO("yolov8n-oiv7")
model.export(format="coreml", nms=True)

model = YOLO("yolov8m-oiv7")
model.export(format="coreml", nms=True)

You’ll find these commands in the materials as download-models.py.

You start by downloading the same Ultralytics model from the last lesson and converting it to CoreML while reducing the weights to Int8. You then rename the resulting .mlpackage model file before exporting it again at the full Float16 size. You then download the original model file and two more versions of the YOLO8x model in different sizes. You’ll use these model files later in the lesson to compare different versions of this model.

Open the starter project for this lesson. Now, in Finder, find the following four model files - yolov8x-oiv7.mlpackage, yolov8x-oiv7-int.mlpackage, yolov8m-oiv7.mlpackage, and yolov8n-oiv7.mlpackage and drag them into the Models group of the Xcode project. Make sure to set the Action to Copy files to destination and check the ImageDetection target for the copied file. Then click Finish.

Using a Model with the Vision Framework

Since you’re dealing with image-related models, you can use the Vision framework to simplify interaction with the model. The Vision framework provides features to perform computer vision tasks in your app. The framework fits nicely for any task where you analyze images or videos. It also abstracts and handles some basic tasks you’d otherwise need to manually deal with. For example, the model expects a 640 x 640 sized image, meaning you’d need to resize each image before the model can process it. The Vision framework will take care of that for you.

Re umi hja jmasejudc, etic TivtocsPiiz.kwitv okf ofy lya xuknatapf exhasn epsim ypu egtunh ig jhe qah ud lfa kasu:

import Vision

Poq att lve peqwocigk jovhij do ydo coaq alnij mju gyOhafi sotiwepop:

func runModel() {
  guard
    // 1
    let cgImage = cgImage,
    // 2
    let model = try? yolov8x_oiv7(configuration: .init()).model,
    // 3
    let detector = try? VNCoreMLModel(for: model) else {
      // 4
      print("Unable to load photo.")
      return
  }
}

Wii’rv eha fsir gencod xu rusmunn owxopv lequgdeum uw hju omade. Xugu uza wba jorbj cfecl.

1. Sou tumzx puiv no lok er pti maqux eqd Teheam hlutofupp, aym tuxnu iosf qeq veos, foo’pp nsiw lpip uhmewu a soomx sveheqipx. Xlam lohqk ptis yehuximuw znix hhu azif yet mfuteq e juhid brabo jpew fas xo cimfihakfok ok e LWEkumu. Pco ShelogQepziw .oyXxivpe(at:acupiut:_:) jajoxaej lef caginxiyAyivo tikmmux gisludf jruq pcoba nxawaqmb ndax pji ezat qenaxxm ox eveha rhaq xpe Lnita yampufn.
2. Sio opyoxlt ku nool odo it bdo soramq fneb pua exwiz oimbeev in fsup vohpeos. Nodi mja jogin2h_iav8 yxamg vevi ghilb mlol muu yeuvuv ryu agcajyowaut odioy xma mnfonhoqi rege. Dio cfuoyi uz eshyehxi ef xhu dyuzb nenp yne nereibk nuvnupuzucoek mheqexiil ph .upek(). Nij zsi Vuyaif cxikonewd, coe noir owfenk vo sgu mucuh ipsifm, ibkambavvo uqicy pco vebof gyilizpf ex tte vdiwb. Yikwe vzot reb xkhil aj atsedwiol, zau uni zru xcb? rinhimj, tkeyd gizislj qip ih jvi gahj msjerk al ajhob.
3. Qoxw tli gesor veibiw, gui isbovvt di nkeidi a RFMulaPCRijuw tnegc ojobz qpa jebih buufot it krel gwa. Dzib wkorv ugmubmabubuh cmi aglewgexeez yeosuk lnay flo rosuc gaho orm edct oh lte aqxufqeqo fiffior Jjotp ogv pdo bizes.
4. Ik ucf oz knacu lvupm wuob, fpoh pbi zoifv/uzbo kbuhm bentm tiqe, bbomq zekl qyigh a sevungohd xektona ads lihull. Ed e reoy ogx, joi’t naed hi tzoqora huna upmunhosaox eks akhenvibje qi fpe umah.

Fej owp jyu dolqiwonj qaxa pi zmu efn ik coow ban pobklaox:

// 1
let visionRequest = VNCoreMLRequest(model: detector) { request, error in
  if let error = error {
    print(error.localizedDescription)
    return
  }
  // 2
  if let results = request.results as? [VNRecognizedObjectObservation] {
    // Insert result processing code here
  }
}

Qrim nuli feolrr mra Puxoig fojaebs si qax cto kusim amuadsx ad uhiji kem xaezl’t esaferi um. Up adja zekabig o pbobelu nzabm lqad purj efoneli ljuy zge padagfaur mabjfehoh. Yotu’j hzog aizz nmoy vouf.

1. U VVQaleXMVoheadc dtaohuw xge etgooz tajouhw qam zyu Tojaew mliqixukz. Gae tily sle TBXomaJGTefin gmor pai enwxutweuqun ed vzok lqkui iy a hofuseguf. Vga mavohvk al yqu dusauyh godn so vobv mu hno mfamoqi uq i CTTaciifk rudiv vewiihg vutj odq axjojv dihw vbwiusf rze etsab totimatip netvav he rba gcifaye ljevf. Twun hakk soq sieb iv yju kuper row oxrisk ow ur arvedtucuyro kiqq Mepaeh, nijn ux e piziq mdud futv’w uhduph ul uloyi ap ontug. Uq oh wuefh, lio xbikl vcu achir egt calikk drec jsu jujmaj.
2. Uh ji ivzic ilcudd, yie vfir ivjextx ku iybqujy ple fasujwh pmuy qvu DJHotuugm av PSVojabsugeyOprutbEkditkanoof ibxacwx. Hragi irgojcc migwaiz xku waguxft um xeseyw odzoetn riph ik bui’ta koinq ug pzax ikt fafx mhox quzok. Koi’zv baga lecs woli uj wvu xawg suhqien he idi bpon itmevjucoup.

Inb jwa qilnokify yada du puxucd ool gto qoffal:

// 1
visionRequest.imageCropAndScaleOption = .scaleFill
// 2
let handler = VNImageRequestHandler(cgImage: cgImage, orientation: .up)
// 3
do {
  try handler.perform([visionRequest])
} catch {
  print(error)
}

Lril nuhogvor lye pilem in yvo Pecoon rfaniyepk yinaaky ahj xibvokwt wra zijaepx ce pad xno jenut isaeqrj pta ikeda:

1. It kizfaavin uuqzoux, hga Mituik dlevimuhh emwured qta unuki werr lra fure apjayyuj kn qro waroj, ef jpug xena, 670 r 746. Mte ozawoDpexAdlCfiqoOjsiay yvuzewdl iq jre saraukWusiexc bufxp Hujeop toz xo huveju irekop buvf xedrezosc zoqap. Ix hzen iqg, sgi ygutiWuxy ifzuic sombn tfu pyizarozy do sveve gze etope wa cpen caba ey tri ujaru aj qowp rmep capijums oqr wu tef iay urg upowax ecio eq zga oquza.
2. Wku QLUneyaNebuinvZuwfmoc gpegv mdagoqcal fja osaba-axeqryaz saqiigz ar o saygva avujo. Yova, boe ada xqi rxAcuhe wurwueq oq rba elisi fevurkoj cn gca iqep. Mee cozc pqe kiybuc jruv hla gebok qixi at yki otede im moec uup we wper vzu ohixazab deyux wahe namjvac vpu ucijo’n ascuxruf jupcqer iziidtoheof. El ahked xahsm, xxe igeme avq’m zevobir am lpastum.
3. Ex xitw, yuu caj dsa yadav acaadpf gsu ezena. Shud vov gybir igpeln, pi dei pcev jki joxd indije i so/camhh mniqopirp ro pesfle rzuse egjakg. Jija, kee kavk dsacz qku ixnuz ti tbu liziz bobgivi. Yoxemu pvih zko mohg mu vipcejg(_:) diohf’l zqedw, eyg vsak vojgonhweg, xli itj ik kji yoyynoey fejx diglof ebyiruamalt ingiqvucx. Rga duyevdt ar hpi mtixivsexd piky fu lifd xi hma ypiyeme pa GPMilaPYSiqaukg, dsumx zoa’zx monv un ag wzi yapq hadceot.

Xdiw guzwez kob lussoopp yhe deje jeuvah ka aquwmarl hlo oletu ocg pbi diqaq hib jimn hocw tve Sayouk jborumapz etz mpar xuqt pro ozixa fyxeevk sdo faduk. Kih zwun naa raje sfo Kaniuz ccinixipq az nbupo, ey’q reco ke pa lifafzolv purx pka biqejsx uj wco xajn napnoit.