WWDC25 Vision Updates

Apple’s WWDC 2025 (June 2025) introduced significant updates to the Vision framework relevant to document scanning and receipt processing. These APIs are available in iOS 19+, macOS 26+, iPadOS 19+, visionOS 3+.

New APIs

RecognizeDocumentsRequest

See recognizedocumentsrequest for full documentation.

• Structured document reading: tables, lists, paragraphs, barcodes
• Data Detection: email, phone, URLs, dates, currency, addresses
• 26 languages, all on-device
• Transformative for receipt scanning (tables → line items)

DetectLensSmudgeRequest

• Detects smudged camera lenses
• Returns SmudgeObservation with confidence 0-1
• Use to reject low-quality captures before OCR
• Helpful for user-facing receipt camera flows

Updated Hand Pose Detection

• Improved model for hand pose tracking
• Not directly receipt-related

Implications for Receipt Scanning

Before WWDC25	After WWDC25
OCR returns raw text + bounding boxes	OCR can return structured tables and detected data types
Must build own table extraction	Tables come as 2D cell arrays from the API
Must implement own date/currency detection	DataDetection framework identifies them automatically
No image quality feedback	LensSmudgeRequest rejects bad captures

Related Pages

• recognizedocumentsrequest — Detailed API documentation
• apple-vision-framework — Parent framework
• ios-receipt-scanning — Comprehensive receipt scanning guide