[Knowledge of Visual Inspection] A discussion about what AI is doing, I wonder.

If you have something you want to find, please feel free to contact Sky Logic!

When talking with customers, I sometimes feel that they believe AI (deep learning) automatically determines what is OK or NG. This is a subtle nuance, and it's not something that needs to be corrected in the course of actual conversation, but today I want to clear up that confusion by writing this article. Now, regarding the confusion mentioned above, the actual difference is that it is humans who decide what is considered OK and what is NG in the software. This leads to the question, "So what does AI do then...?" To put it simply, what our AI does is "find what it has been taught," and fundamentally, that’s all there is to it. This function is generally referred to as "object detection." In "object detection," it literally detects "objects" within an image.

Product Inquiry

basic information

If you teach it to detect screws as the object to be detected, it will learn to find screws, and if you teach it to detect scratches, it will learn to find scratches. At this point, if you teach it that certain items are "good products" and others are "defective products," it will also learn to distinguish between "good" and "defective" products. However, there is no judgment of "good" or "bad" involved; it simply means that it has found an object labeled as "defective." Once an object is found, it counts how many there are. This allows us to understand things like there being 10 screws in the image, 1 scratch, or 5 processing holes, as well as whether an object is facing right or left. This ability to determine how many of each object there are is the role of deep learning (commonly referred to as AI). At this stage, simply counting the number of objects does not indicate whether they are acceptable or not. It is only when humans provide acceptable limits for the counted quantities that OK/NG judgments can be made. If you view AI as a simple tool for finding objects, it may become clearer how to utilize it. If there is something you would like to "find," please feel free to contact us.