Thanks, it renders a bounding box around each word that is recognized, you can then click on the box to fill it. It's very accurate if you give it an image with visible structured text.