How to detect documents using VNDocumentCameraViewController

Swift version: 5.6

Paul Hudson @twostraws June 3rd 2019

iOS 13.0 introduced a new micro-framework called VisionKit, which is specifically designed to make it possible to scan documents like Notes does.

You can then Vision OCR to scan the text if you want, but by default VNDocumentCameraViewController just gives you images of each page.

To get started:

Import VisionKit.
Make some type (such as your view controller) conform to the VNDocumentCameraViewControllerDelegate protocol so you can handle delegate callbacks.
Create and present an instance of VNDocumentCameraViewController, setting its delegate property to whatever should be notified when a scan completes.
Present the document scanner as normal, then wait for feedback.

So, something like this:

let vc = VNDocumentCameraViewController()
vc.delegate = self
present(vc, animated: true)

Once the scan completes your delegate will get called with the document, like this:

func documentCameraViewController(_ controller: VNDocumentCameraViewController, didFinishWith scan: VNDocumentCameraScan) {
    print("Found \(scan.pageCount)")

    for i in 0 ..< scan.pageCount {
        let img = scan.imageOfPage(at: i)
        // ... your code here
    }
}

The result of imageOfPage(at:) is a UIImage, so you’ll need to replace “your code here” with whatever you want to do with your images.