I attempted to OCR a PDF whose first page is normal text pdf and the following pages are scanned.  Oddly the image pages returned as OCR'd text, but the text page was returned as an image without OCR.

How can I process this correctly?