Hello to whom this may concern,
I have made a form processing model to scrape data items from a pdf invoice using the form processing model. For my model, while creating the model the confidence of 100 % was displayed, and no values where skipped, however when I parse those values in my flow values are being skipped in no specific order and this is occurring with multiple pdfs that the model was specifically trained to. My second approach was to extract the whole table as a whole instead of column-wise and this extraction as a table was rather more inconsistent then column-wise extraction values where being skipped characters were being rude as different values and etc.
Any Solutions to get rid of the inconsistency since whole data items are being skipped, just a reminder the invoices for these models are high quality.
Assuming this is an AI Builder issue for fetching data from table? Please advise your opinion and potential solution.
Thanks in Advance.
Thanks @AaiyanVirji for verifying.
In your flow, have you selected the table items that finish by value? In my following example I have a table like this:
Item | Quantity | Rate | Amount |
Monthly service fee | 1 | $40.00 | $40.00 |
Add-on services | 2 | $6.00 | $12.00 |
So to iterate each item in flow I select, table_0 Item value, table_0 Quantity value, table_0 Rate value and table_0 Amount value.
When I select one of the items, the action where I add any of these values gets added automatically to an Apply to each loop.
I have done that, and the values are coming up just the same when I test the model but, when I try to parse the data it fails to get all values and all values have been enclosed by rectangle meaning they are being fetched by the model hence not sure why the values aren't all being parsed.
Hi @AaiyanVirji,
Thanks for reaching out.
It can be that not all elements on the table are being recognized correctly. The fastest way to see this visually is to: