Visionkit can lift a subject. But the bounding rectangle is always returning x,y,width,height values as 0,0,0,0

Question

Created Dec ’23

Replies 1

Boosts 0

Participants 2

In our app, we needed to use visionkit framework to lift up the subject from an image and crop it. Here is the piece of code:

if #available(iOS 17.0, *) {
            let analyzer = ImageAnalyzer()
            let analysis = try? await analyzer.analyze(image, configuration: self.visionKitConfiguration)
            let interaction = ImageAnalysisInteraction()
            interaction.analysis = analysis
            interaction.preferredInteractionTypes = [.automatic]
            guard let subject = await interaction.subjects.first else{
                return image
            }
            let s = await interaction.subjects
            print(s.first?.bounds)
            guard let cropped = try? await subject.image else { return image }
            return cropped
        }

But the s.first?.bounds always returns a cgrect with all 0 values. Is there any other way to get the position of the cropped subject? I need the position in the image from where the subject was cropped. Can anyone help?

Boost

Answer 1

interferon OP

May ’24

This happens if the interaction doesn't have an associated view. I think this should probably emit some kind of warning, but instead it just returns (0,0,0,0) as the bounds.

To fix it, you need someImageView.addInteraction(interaction) before you access the subjects array.

0