Videos

Learn how to use the Anomaly Detector module, which identifies novel defects by training on defect-free images without annotations. This tutorial guides you step by step, from setup to calibration, for efficient defect detection.

The Detector module is ideal for identifying and classifying objects or defects, verifying components, and sorting products. This tutorial covers everything from image selection and annotation to setting confidence thresholds.

The Surface Detection module specializes in inspecting surfaces for manufacturing flaws and cosmetic defects. This tutorial covers key steps, including image selection, defect annotation, setting evaluation criteria, and fine-tuning results.

How do you detect deep scratches on aluminum tubes while ignoring shallow, cosmetic ones? With PEKAT VISION, you simply teach the system what matters — it learns to focus on critical defects and skip the rest.

We recently collaborated with Arimation Robotics to create a deep learning solution that sorts over 6,000 PCB boards into 20 categories by precious metal content—an otherwise impossible task due to its volume and complexity.

We demonstrated an automated gearbox assembly with integrated quality inspection. Gear wheels were inspected for defects using PEKAT VISION, while the Impact software measured diameters—all powered by the MX-G2000 Vision Processor.

Using the Detector and Classifier modules, PEKAT VISION verifies the order, orientation, and completeness of chair components before packaging. Watch our short demonstration.

Petr Šmíd, the founder and CTO of PEKAT VISION explains the differences between deep learning and traditional algorithmic methods in quality inspection.