Manufacture of printed goods including commercial printing and packaging printing is made up of craftsmen's skills and complicated quality control. The value of printing is appraised when those products display at the place of public and when they are taken in hand. In order to control their quality, it is necessary to identify and sensitize delicate colors close to the human eye. This product provides a mechanism to easily install advanced image recognition processing of deep learning technology specialized in printing manufacturing on site.
This system provides Web API designed to be easy to use and SDK and library assuming to implement into the device.
Workers in the field with the print know-how can label data (annotation) themselves such as image classification.
It's new learning methods that minimize risks from the viewpoint of leakage of defects by the optimization algorithm.
In order to solve the problem that this pre-training is difficult, it provides a mechanism to construct a fast image processing and flexible system.
Until now, we have confirmed and classified the defect images detected by the inspection machine after printing one by one with human eyes. From now on, AI automatically classifies the defect type and classifies it.