Use OCR to identify the coordinates of numerals on a technical drawing PDF

---

Description:

Project Overview I have technical drawings of cabinet panels in PDF format. These drawings contain number labels for which I need toe extract their precise positions. However, the numbers are part of an image (not selectable text) and cannot be extracted using regular text-based OCR methods. Challenges & Requirements The numbers are printed over thin vertical and horizontal lines, making extraction difficult. Font size and font family can be adjusted if needed to improve accuracy. Some form of image manipulation may be required to remove or lighten background lines for better recognition. The solution should achieve at least 97%+ accuracy in identifying number positions. Delivery format: Preferred: JavaScript (Node.js) Alternatively: Any language that can be deployed as a Google Cloud Function, where a PDF is sent via HTTP and the function returns the coordinates of detected numbers. Success Criteria The solution must consistently and reliably detect the numbers with a minimum 97% accuracy rate. If the system meets this threshold, you will receive full payment. If the solution does not meet this threshold, the job will be considered incomplete. While I won’t leave negative feedback for an honest attempt, I can only pay for a successful outcome. Additional Details I can provide multiple sample PDFs to test against. A brief video chat may be required to clarify project requirements and ensure alignment before starting. If you believe you have the expertise to tackle this challenge, please apply with details on your approach and relevant experience. Looking forward to working with you!

Tags: Computer Vision, OCR Algorithm, OCR Software, PDF

Keyword: Cloud Computing

Job Type: Fixed price

Price: $1.0