Vision is one of the many exciting new frameworks introduced at WWDC 17. Enhanced Text detection is among the many new features in Vision.framework. But since Apple hasn't provided any demo code on the Vision text detection, here is a quick sample project.
This demo app:
- takes a live video feed, detects text regions (VNTextObservation) and draws a green box around each region
- draws blue boxes around each character box detected (VNRectangleObservation)
Running this app requires at least Xcode 9 beta 1 and a physical device running iOS 11 beta 1.
There is no OCR framework in the iOS SDK still. As a result, a seperate OCR engine is needed to recognize detected text content.