AI Drawing Number Auditor for Engineering Packages
Checks engineering drawing files against a drawing index using AI vision, so misnamed or misplaced drawings are flagged before a package is filed.
Problem / Challenge
Industrial drawing packages, including P&IDs, termination sheets and vendor drawings, must be checked against a drawing index so misnamed or misplaced files are caught. Doing this by opening every PDF is slow, repetitive quality-control work.
Solution
A Python tool reads the drawing index from a spreadsheet and converts each PDF page to an image. It isolates the title block and reads the drawing number with the OpenAI Vision API. It then fuzzy-matches that number against the file name and the expected entry, sorts each file into a result folder and logs the outcome to CSV.
How it works
- Load drawing index (Excel)
- Convert PDF pages to images
- Crop the title block
- Extract the drawing number with AI vision
- Fuzzy-match against file name and index
- File into a result folder and log to CSV
About this project
This project is from the professional portfolio of Muhammad Umer, Founder & CEO of EaseWorkflow. It is shown as an example of his automation work, not as an EaseWorkflow client engagement. Client and identifying details have been removed.
Results / Outcome
- Designed to reduce manual drawing-package checking.
- Designed to lower the chance of a wrong drawing being filed.
- Produces a CSV log of every check.