Issue with Document Intelligence Custom Model for BOE Extraction and Alternative Solutions

Question

I'm facing issues with a custom Document Intelligence model configured to extract fields from Bills of Entry (BOEs) presented in tables. Specifically, the model:

• Fails to extract all fields accurately from BOEs in table format.
• Does not label all fields correctly when processing multiple pages of BOEs.
• (my data is in pdf extracting in excel, trained the model working fine with single page pdf but not with multiple pages)

Could you please provide guidance on optimizing the model settings or any specific considerations for handling multi-page documents in Document Intelligence? Additionally, I'm interested in learning about alternative services or workarounds that could achieve similar functionality.

Answer 1

Hi there Niket Kumar Singh

Thanks for using QandA platfrom

It sounds like the issues with your custom Document Intelligence model for extracting fields from multi-page BOEs.

make sure the model is trained with a diverse dataset including multi-page documents. and Check if there are any specific settings for handling multi-page documents in the model configuration.

• Split the PDF into individual pages before processing and then merge the extracted data.
• Use a pre-processing step to label and segment data across pages consistently.

for the alternative solution i guess form recogniser its Good for extracting structured data from forms and tables.

If this helps kindly accept the answer thanks much.