Extract Table data from Technical Drawings Reliably

Hi everyone,

If you have coding skills (preferably in Python), this might interest you.

It may even be possible to solve without coding, but I haven’t found a reliable no-code approach yet.

The challenge:

I have attached two types of technical drawings:

The PDF drawing contains one simple table on the first page.
The image drawing contains two tables, but only the one titled “Parts List” needs to be extracted.

The task is to reliably extract all information from the correct table. The solution must be bulletproof, with no room for failure.

Scope of the system:

Input will always be PDF files (not images).
The target table will always be titled “Parts List”, containing the columns like:

- ITEM

- QTY

- PART NUMBER

- DESCRIPTION

- Maybe even more columns named something else.

The table’s position in the PDF can vary: right side, left side, or even on a separate page.
OCR is not needed, since I can already extract text from PDFs.
I’ve tried GPT-4.1 for image recognition, but when processing A3 sized PDFs, it struggles - tables appear too small and errors occur.

My attempts so far:

I’ve experimented with Python using tools like Camelot and Pdfplumber, together with the “Excel Viewer” extension in VS Code. See the python image (3) attached.
The code correctly extracts all table data (see image 4 for output), but the result is unstructured.
The next step is to transform the extracted data into a structured format (e.g., transform into a spreadsheet) so it can be used effectively.

My question