Extracting content from XLSX - options from Azure Form Recognizer

Question

Hello Team,

I hope this thread would be the point of discussion on recent advancements in Azure Form Recognizer, I want to read excel files using form recognizer - I tried using layout model - it was able to get all cells data as individual paragraph elements - still no concrete way to establish the boundary in terms of (no of rows and no of columns) - Similar to pandas excel file reader - does form recognizer supports any specific methods that improves the accuracy in extracting EXCEL file handling?

Answer 1

@Venkatesan,Sangeetha welcome to the Microsoft Q&A community.

Azure Form Recognizer does support extracting data from Excel files, but it seems like the layout model isn't providing the structured tabular format you're looking for. While the documentation mentions Excel support, users have reported challenges in accurately defining row and column boundaries.

One workaround is using the OCR API, which supports Microsoft Word, Excel, PowerPoint, and HTML file formats. The Prebuilt-layout 2023-10-31-preview model has been mentioned as having improved table extraction capabilities, but its full implementation status isn't clear yet.

Additionally, Microsoft has been working on enhanced table extraction to improve accuracy in detecting tabular structures, including handling merged cells, nested tables, and complex layouts. If you're looking for a more structured output similar to Pandas' Excel reader, you might need to preprocess the extracted data using Python to reconstruct the table format.

I hope these helps. Let me know if you have any further questions or need additional assistance.

Also if these answers your query, do click the "Upvote" and click "Accept the answer" of which might be beneficial to other community members reading this thread.