Coordinates of detections

#1
by mileseverett - opened

Model knows where it has detected text regions, as shown in the output image. But would be a good output to also contain bounding box of this information as this could then be used for structure.

Great idea @mileseverett I had been wondering if it could collapse into a document box model where it regenerates an exportable friendly table or similar.

With some of the image to OCR if you put text back onto the coordinates but get text size wrong, it is often difficult to read or save locally. I'd like to find a happy medium to make sure output text is always exportable or convenient to structure as table or document outline structured info.

I'll take a look at reworking this to somehow get more usable exportable media to be able more easily use it on IoT devices like phones and ipads where carrying it around with you makes alot of sense.

Thanks for feedback!

--Aaron

awacke1 changed discussion status to closed

Sign up or log in to comment