Hello!

 

I am unable to find a way in which to train line items that works well!

Currently, in the training batches, after the initial recognition I delete the table and then create it by myself, manually. This seems to help the recognition results, but not by a large margin. I strongly believe there is a better way of doing this than the one I am currently using. Does anyone have any idea of which is the best way to train the extraction of lineitems?