Assessing and deﬁning the core problem.
Installing hardware to capture high resolution image data.
Labelling image data to determine quality thresholds.
Training and ﬁne-tuning models to detect defects and trigger actions.
Deploying the trained model with continual monitoring to account for process variation.