Case studies

Each part manufactured requires weld inspections to identify defective or missing welds. Ensuring the quality of welds is essential for guaranteeing the parts safety. Up to this point, humans have been responsible for conducting these inspections.

Canam enlisted Vooban to create and implement an automated inspection system for steel joists that run on conveyors.

Vooban has developed a complete software solution for detecting faulty welds. It includes an app that allows users to visualize these errors easily.

The solution includes an AI-powered machine vision system that operates in real time using live camera feeds. It can scan up to 5 stacked joists at once and detect up to 3 different types of welds.

The model also includes an app that makes it easier for users to visualize errors. It can notify inspectors when a defect is detected and allows them to easily tag images with poor detection scores. These images will then be used for model retraining and improvement.

This technology allows Canam to affirm that they perform the highest quality control in the industry. A single missing weld on a joist could pose a danger if used in construction. However, with the near-perfect performance of the AI system, a Canam product is a certified safe product.

The system achieves nearly 100% accuracy, drastically reducing manual inspection time. By automating the inspector’s role, production capacity has increased since employees’ time can now be redirected to welding. Across all production lines, the project has led to substantial efficiency gains.

Given the current labor shortage, particularly among welders, this project gives Canam a significant competitive edge in the market.