vision-framework

On-device OCR and barcode scanning in iOS apps can be implemented using the Vision and VisionKit frameworks to extract text and capture data without requiring a network connection. This configuration exposes these features through the modern ImageProcessingRequest pattern and DataScannerViewController for live camera feeds.

The modern Vision API uses the ImageProcessingRequest pattern available on iOS 16+, while legacy VNRequest patterns provide backward compatibility for older iOS deployments. Both approaches handle computer vision tasks like face detection and segmentation, but the modern path is required for advanced VisionKit features.

VisionKit DataScannerViewController for live camera scanning requires iOS 16 or later and does not work on older iOS versions. For older deployments, you must use the legacy VNRequest path to achieve similar computer vision functionality without the modern VisionKit interface.

You can run custom Core ML models alongside Vision to perform image segmentation and instance segmentation tasks. This integration allows you to combine built-in Vision capabilities with specialized models for selective visual effects and object tracking within your iOS application workflow.

Adding face detection with landmarks to an iOS application is best handled through the Vision framework to capture pose cues for interactive experiences. This approach coordinates Vision requests alongside OCR and barcode scanning to consolidate multiple computer vision features into a single reliable module.