WWDC25: Read documents using the Vision framework | Apple

Apple's Vision framework introduces two new APIs at WWDC25: RecognizeDocumentsRequest for extracting structured information from documents including tables, lists, and data detection, and DetectLensSmudgeRequest for identifying photos taken with smudged camera lenses. The framework also updates its hand pose detection model with improved accuracy and reduced memory usage. These APIs run entirely on-device across iOS, macOS, iPadOS, tvOS, and VisionOS, enabling developers to build apps that can parse complex document structures, detect contact information, and ensure photo quality.