Turn Key Defect Detection for Tubulars

Client requested a vision system capable of detecting defects on premium oil and gas tubulars. A turn key, patented system was developed where a robotic arm is used to capture ultra high-resolution images of box and pin threads. We also developed a proprietary Deep Learning model that is capable of being retrained by an inspector in the field on novel defects, with very little training data and cycle time.

The inspection sequence is performed in a virtual environment beforehand.

The result is a concatenated 4K image with submicron level resolution with no image blur or vibrational artifacts.

A proprietary Machine Learning model is used to identify defects on the threads.

Some samples of defects identified

Previous
Previous

Computer Vision & Deep Learning

Next
Next

Slimhole Imaging