Deep learning methods in machine vision

Assigns predefined classes to objects or image regions, for example OK/NOK or type A, B, C.

Used for e. g. quality control, sortation, or identification of material.

Locates and classifies multiple objects simultaneously, even when they overlap.

Used for e. g. pick-and-place, assembly inspection, and robot vision.

Detects unknown or rare defects automatically without prior defect examples.

Used for e. g. surface inspection, leak testing, defect detection.

Assigns a class to each pixel and separates objects precisely.

Used for e. g. texture recognition, part separation, and pose estimation.

Detects individual object instances of a class, even when overlapping.

Used for e. g. multi-object detection and packaging inspection.

Learns relevant edge structures automatically for subsequent evaluation.

Used for e. g. gap measurement, contour tracking, and joining processes.