Company Profile

Keye Technology relies on the Keye Laboratory of Pattern Recognition of the University of Science and Technology of China, and has a high-level doctoral R&D team covering the entire technology chain of optics, mechanics, electricity, computing and software.

Through the project practice in the fields of packaging containers, 3C, medicine, textile, food, etc., the KVS-AI software and hardware platform has been created, so that AI technology can be transformed into real productivity. We provide visual imaging, algorithm computing power and automatic control AI system solutions for various sub-industries.

Inspection Principle

The packaging materials of glass bottles are becoming more and more diversified, and the overall requirements for packaging manufacturers are also constantly increasing. In particular, the requirements for the geometric size and appearance quality of glass bottles are becoming more stringent.

Due to the high temperature of the glass bottle production environment, more than 80% of the inspection processes currently used by most glass production companies are manual. The inspection standards are not uniform, the efficiency is low, it is highly influenced by subjectivity, and it is prone to errors. To change these problems, automated machine vision inspection equipment needs to be used to replace manual inspection.

Defects Type

Inspection Details

This solution is designed based on the actual workpiece conditions on site, and visually inspects the geometric dimensions and appearance defects of transparent glass bottles to replace manual inspection work. The inspection system integrates visual optical hardware, AI artificial intelligence inspection system software, defective product rejection and counting devices. The inspection equipment is connected and installed. The equipment can operate 7*24 hours a day, greatly improving production efficiency.

Equipment measurement size and appearance inspection adopt photo identification inspection. One picture is collected for size inspection, and more than 20 pictures can be collected for one glass bottle for appearance defect inspection. The specific number of collections will be determined according to the actual inspection situation. After the system collects sample images, it determines good and defective products through comparison and analysis with the image processing software and standardized samples. The defective products are detected and automatically eliminated online. After the finished product is inspected, the counting function can be realized, and the counting accuracy is 99.9%.