Tenaris, a leading industrial pipe production company, proposed this project to automate their manual straightness checks. Team i18 envisioned a machine learning solution. We constructed a vision system using a Triton 24.5 MP Model sensor to capture pipes in production settings and feed them into a machine learning model. This model, built from Tenaris pipe images, identifies flawed pipes based on a 3-5 second video. The Team created a web-hosted user interface, allowing factory managers to view the model's diagnosis and provide their validation with a few simple clicks. While currently hosted on Google Drive, this system is transferable to desktops, company servers, or OneDrives with minimal coding changes. The design supports expanding the model's dataset, allowing the 95-98% accuracy performance to further improve. Our final product combines hardware design, data science principles, python code, and user interface design to provide an automated solution for real-world quality control standards. We are proud to combine our collective expertise and work ethic to contribute to the growing applications of machine learning automation.

Team Members

Mona McElroy
Ben Braun
Jack Krieger
Humza Syed
Aaron Chang