Hi all,
I have trainined a custom document model in AI builder, the model its a document that has some defined fields where people handwrite to complete information.
When I train the model to read a field that should be a 5 digit number, if the last number is "1" it often gets missed. E.g 34534 gets read ok, but 34531 gets read as 3453.
This happens even when I tran the model, I select the entire box where text may appear, but the model only wants to select the text it thinks is important in that area.
Any ideas why this is happening?
Thanks,
Hi @sossie071 ,
It can happen if the character is not fully visible or overlaps with other characters. You can try to train your model with samples of higher quality.
However, at predict time, the model may not recognize such characters anyway. In that case, you can set a condition in your flow that checks the length/format of your number and store the document elsewhere for a later manual review if the check fails.