Overview
Back in 2019, our Tech Lab at 3techno Digital took on a bold challenge: building an OCR utility to extract structured data from multi-language National ID cards, specifically in English and Urdu. The idea was to enable automatic data capture from ID photos—usable across systems like HCM onboarding, visitor management, and identity verification.

Though this initiative was later shelved, it became a formative part of our journey in AI, automation, and enterprise tech experimentation.

What We Set Out to Do

• Recognize text from National IDs using real-time images
• Support dual-language extraction (English + Urdu)
• Structure the extracted data for seamless integration into systems
• Reduce manual data entry in admin-heavy workflows

What We Explored

• Open-source and commercial OCR tools: Tesseract, Google Vision, etc.
• Preprocessing techniques: noise filtering, image correction, perspective flattening
• Custom field-level mapping for key data like Name, CNIC, DOB, and Address
• Edge testing with real-world input: blurry scans, low-light photos, mixed fonts

Why We Shelved It
While early tests showed promising accuracy, the complexities of Urdu script handling, mixed-language formatting, and production-readiness required more resources than we could allocate at the time. Rather than rush an incomplete product, we made the strategic decision to shelve the project.

The Value It Gave Us
This R&D initiative, even without commercialization, gave us:

• A deep understanding of document processing challenges
• Practical skills in OCR benchmarking and pipeline building
• Internal tooling that later informed other AI research tracks
• A stronger culture of hands-on experimentation and rapid prototyping

3techno Tech Lab | Archived R&D (2019)
Every project teaches us something. Some even shape our future.
www.3techno.com | 📧 [email protected]