Why ever?

We have received many evaluations from users who have introduced FlexInspector or are evaluating it on their actual equipment, and many of them have said, "This is the only way to go.
I personally feel that FlexInspector's inspection method is the perfect form of automated visual inspection using image processing.
The first point is simplicity.
By simply imaging multiple good parts, good parts are defined automatically and objectively, and any defects can be detected as "differences from good parts".
With this method, there is no need to be aware of what defects may arise at all.
Many people will find this alone appealing.
However, those who have actually tried it seem to appreciate it in a different way.
The "detection capability" and the "correlation with visual inspection" are amazing.
As for the detection capability, it is thought that aiming for "more than this" will cause a decrease in yield because the line is drawn at the very edge of product variation. In fact, if the line is drawn at the "average ±3σ" level, the yield will be considerably lower.
We have introduced our original "trick" to obtain an extremely high correlation with visual inspection.
This "trick" is so simple that it can be easily understood.
Why have we been doing this?" is one of the reasons for the high evaluation.