Industrial Computer Vision with AI: Defect Detection and Visual Inspection Training Course

Introduction to Industrial Computer Vision

• Perspectives on machine vision systems within manufacturing sectors
• Common defect types: cracks, scratches, misalignments, and missing parts
• AI-driven methods versus traditional rule-based visual inspection

Image Acquisition and Preprocessing

• Camera variants and optimal image capture configurations
• Techniques for noise reduction, contrast improvement, and data normalisation
• Applying data augmentation to ensure model robustness

Object Detection and Segmentation Strategies

• Conventional techniques such as thresholding, edge detection, and contour analysis
• Deep learning frameworks: CNNs, U-Net, and YOLO
• Selecting between detection, classification, and segmentation approaches

Developing Defect Detection Models

• Structuring annotated datasets for training
• Training classifiers and segmentation models for defects
• Assessing model performance through precision, recall, and F1-score

Industrial Deployment Considerations

• Hardware specifications: GPUs, edge devices, and industrial PCs
• Architecting real-time inspection pipelines
• Integration with PLCs and broader factory automation systems

Optimisation and Maintenance

• Adapting to variable lighting and evolving production conditions
• Implementing model retraining and continuous learning cycles
• Integrating alerting systems, logging, and QA reporting workflows

Real-World Case Studies and Applications

• Detecting defects in automotive assembly and welding processes
• Surface quality inspection for electronics and semiconductor components
• Verifying labels and packaging in pharmaceutical and food industries

Recap and Future Directions