How to Fetch table from image

ocrextractrelative works with green and pink boxes. The part inside the green box (called anchor) must not change, it can be e. g. the table header or something else nearby.

And the pink box is around the changing table. If the table would be in this forum, here is what I would use:

image

=> Only the part inside the pink box is send to the OCR API for processing.

PS: For numbers, use OCR engine2.

1 Like