Improve Cognitive Services OCR 2.0 performance to equal the OCR Demo
OCR API 2.0 Sample Code is not as comprehensive, nor accurate as running the OCR Demo on the exact same image. (Tried many!)
It is clear there are differences. The two return different dimensions (corners vs corner, width, height). No Cognitive Services documentation indicates what these differences are.
I suggest the differences should be stated or, better yet, the Sample Code be upgraded to replicate the performance of the Demo.