Quality detection system for filling infusion bottle production

Abstract

The invention discloses a quality detection method, system and device for filling infusion bottle production, and the method comprises the steps: transferring a sample infusion bottle into a quality detection system through a mechanical arm, and obtaining sample image data outputted by a white foreign matter detection station and a black foreign matter detection station, extracting structural features of sample infusion bottles and quality features of the infusion bottles from the sample image data to serve as a first sample set; adding the first sample set into the current sample set, and adjusting parameters of an upper computer identification model; transferring the sample infusion bottles into the quality detection system again to obtain a second sample set, and adding the second sample set into the current sample set until a trained target model is obtained; and detecting the quality of the infusion bottle by using the target model. According to the embodiment of the invention, the capacity of the current sample set for training the model can be simply, automatically and quickly expanded, and more accurate training of the target model can be realized.

Description

technical field [0001] The invention relates to the technical field of product inspection, and more particularly to a quality inspection system for filling infusion bottle production. Background technique [0002] Before large infusion products are finalized and flow into the society, many links such as measurement, positioning, correction, and identification are involved. [0003] In the production process of canned infusion bottles, it is inevitable that unqualified defective products will appear, which not only brings hidden safety hazards, but also may affect product quality. The traditional manual light inspection method is difficult to ensure the reliability and rapidity of the inspection, which causes some substandard products to flow into the market and damages the corporate image. [0004] At present, the market is limited by hardware, and the general image acquisition system has poor color resolution. Although many domestic companies are committed to the research ...

AI Technical Summary

Problems solved by technology

[0003] In the production process of canned infusion bottles, unqualified and defective products will inevitably appear, which not only brings safety hazards, but also may affect product quality

The traditional manual light inspection method is difficult to guarantee the reliability and rapidity of detection, thus causing some unqualified products to flow into the market and causing damage to the corporate image

[0004] At present, the market is limited by hardware, and the general image acquisition system has poor resolution of colors. Although many domestic companies are committed to the research and development of light inspection machines, due to problems such as insufficient mechanical manufacturing technology and poor detection algorithms, the developed light In the process of detecting foreign objects, the detection machine has a high rate of missed detection and false detection, and repeated detection is required, which leads to low detection efficiency

Therefore, the prior art has the technical problems of low detection efficiency and low accuracy

Images