| CPC G06T 7/001 (2013.01) [G06F 18/2148 (2023.01); G06F 18/217 (2023.01); G06F 18/2413 (2023.01)] | 1 Claim |
|
1. A non-transitory computer-readable medium comprising instructions that, when executed by a processor, cause the processor to control a manufacturing execution system that includes a machine vision module and a classification deep learning (DL) model, the system being controlled to perform functions of determining a quality of individual products and retraining the classification deep learning (DL) model trained to inspect products for defects, the functions including:
in response to a shooting trigger, capturing, with the machine vision module, a first image of a first product;
conducting a vision inspection with the machine vision module using the first image and outputting a vision inspection result to the manufacturing execution system;
providing the first image to an edge computer operating the classification DL model pre-trained by a server to determine, based on a set of parameters, whether a product has an acceptable quality based on an image capturing the product and output a DL inspection result to the manufacturing execution system,
wherein the classification DL model is configured to perform:
in response to the first image provided thereto, attempting to determine, based on the first image, whether the first product has the acceptable quality;
determining that whether the first product has the acceptable quality cannot be determined using the first image; and
in response to determining that whether the first product has the acceptable quality cannot be determined using the first image, determining that the first image contains an error image;
in response to determining that the first image contains an error image, adjusting, based on the error image, the set of parameters of the classification DL model on the edge computer;
extracting, from the edge computer, the adjusted set of parameters of the classification DL model; and
transmitting the adjusted set of parameters of the classification DL model to the server via a communication network;
wherein adjusting the set of parameters of the classification DL model on the edge computer is limited to a preset range, and
wherein, to reduce a false defect rate, the manufacturing execution system operates the classification DL model to re-inspect the first image when the machine vision module has determined the first product to be “not good,” and, to improve a detection rate of undetected true defects, the manufacturing execution system operates the classification DL model to re-inspect the first image when the machine vision module has determined the first product to be an “OK” product.
|