Technology stack
- Computer Vision
- OCR
- AI Validation
- AWS
- Desktop Application
About the project
Defender Tubular Services needed to replace manual serial number recording after pipe cleaning at oil rig sites across the eastern United States. We built a computer vision solution that captures, reads, and validates serial numbers on pipe surfaces, then synchronizes data in real time for tracking and reporting.
The challenge
Manual recording produced frequent errors—especially on worn or partially damaged serial numbers—while delays and limited visibility made it hard to maintain accurate pipe data across multiple sites. The solution had to fit existing operational workflows and scale reliably in the field.
Our solution
We developed OCR models trained for real-world pipe surfaces, paired with AI-based validation to detect and correct recognition errors. A desktop application supports on-site usage, with real-time Excel synchronization and AWS cloud deployment. A proof of concept validated feasibility in two weeks before full delivery in eight weeks.
Results & impact
The system achieved 85% recognition accuracy compared to 30% with manual methods, a 53% faster logging process, and a 70% reduction in data entry errors. Defender Tubular gained digitized traceability, real-time maintenance data access, and a scalable AI foundation for future expansion.