Vision-App

iOS app that identifies objects using coreML written in Swift4

Preview

Built with

• Ios 11
• Xcode 9

Requirements

SquezeNet.mlmodel Detects the dominant objects present in an image from a set of 1000 categories such as trees, animals, food, vehicles, person, etc.

import AVFoundation //Work with audiovisual assets, control device cameras, process audio, and configure system audio interactions.
import CoreML //handles Machine-learning
import Vision //handles Face-Object recognition

Features:

• Access Camera

  var captureSession: AVCaptureSession!
  ...
  captureSession = AVCaptureSession()

• Capture Photo and Save it in small preview

  var cameraOutput: AVCapturePhotoOutput!
  ...
  cameraOutput.capturePhoto

• Control camera Flash

  let flashControl = AVCapturePhotoSettings() 
  flashControl.flashMode = .on

• Tries to Identify Object we take a Picture of

  (request: VNRequest)
  results = request.results as? [VNClassificationObservation] //"VNClassificationObservation" observes what was in the image

• Shows accuracy of Object identification

  classification.confidence

• App reads Results outloud using AVSpeechSynthesizer

  var speechSynthesizer = AVSpeechSynthesizer() 

Credits

Devslopes CoreML Tuts

License

Standard MIT License