Seeing what I may have done to cause the attached issue....

lbellipanni

While working through some Grooper labs came across the attached issue.  It appears the coordinates and size for my extractions are off.  Is this something that could have been caused by Image Processing or do I have some setting set incorrectly.  I am running 2.60.0097 

GrooperGuru

The likely scenario, as others have mentioned, is that in your OCR Profile you are using an IP Profile that causes a change or shift in pixel location. Commands like resize, crop, deskew, and border removal cause these shifts. Those commands must be run using an Image Processing activity prior to the OCR activity, or else you WILL have this issue every time. You cannot use these IP Commands during or after OCR or else subsequent Data Instance highlighting will be shifted.

[Deleted User]

I've seen this before and I do believe it's a result of image processing.  It's still recognizing the OCR text, but the new image (after whatever image processing steps were performed) no longer visually matches.

There's a mention of this being fixed in the 2.60.0079 Release Notes, but maybe it's resurfaced somehow.

• The failure of matching data instances to highlight in the Text Viewer when editing an extraction pattern has been repaired

lbellipanni

Thanks.  First time I have had it happen, so there is something I am about the image processing profile I am using on this particular doc causing it to happen.

[Deleted User]

The best way to find out for sure would be to perform the same activities on the document that you did until that point, minus any image processing.

RandoCalrisian

Are you performing IP after OCR?

lbellipanni

As I look back I probably did Image Processing after OCR.  I started fresh with the document and everything is fine.  Thanks for all the feedback