Computer Vision for Wafer Inspection is the application of image processing and deep learning to automate the visual inspection of semiconductor wafers — detecting defects, particles, pattern anomalies, and process signatures across optical, SEM, and other imaging modalities.

Key Computer Vision Tasks

• Defect Detection: Find defects that deviate from the designed pattern (die-to-die comparison, reference-based).
• Pattern Recognition: Classify defect patterns on wafer maps (systematic vs. random signatures).
• Die-to-Database: Compare captured images against the design layout to find missing or extra features.
• Automatic Defect Review (ADR): Revisit detected defects with higher resolution and classify them.

Why It Matters

• Throughput: CV processes wafer images at production speed (>100 wafers/hour).
• Sensitivity: Modern algorithms detect defects smaller than the imaging resolution using statistical methods.
• Recipe Development: ML-assisted recipe development reduces time to qualify new defect inspection recipes.

Computer Vision for Wafer Inspection is teaching machines to see defects — applying image analysis at production speed to find every anomaly on every wafer.