Challenge
PNK Group aimed to reduce defects in precast concrete manufacturing by improving detection of geometric inaccuracies, misalignments, and surface defects traditionally reliant on error-prone manual inspection.
Solution
We developed an AI-powered vision system using handheld 3D scanners and a mobile multi-camera rig, with machine learning identifying geometric deviations and surface flaws with real-time alerts.
Results
- Automated defect detection replacing manual inspection
- Real-time alerts on geometric and surface defects
- Modular system expected to cut future adaptation costs by 20–30%
Technology
OpenCV · Python · PyTorch · Open3D · Zivid SDK
Powered by DAC Digital