Detection of bottle appearance defects
Bottle appearance defects are a common problem phenomenon in daily production. For manufacturers in pharmaceutical, dairy, wine, condiments, daily chemicals and other industries, due to the increasing requirements for product packaging refinement, quality and continuous mass production, the market has put forward higher standards for the original quality inspection of bottles.
The emergence of machine vision inspection machines based on AI algorithms has become a new tool to help enterprises produce efficiently.
keyetech
Bottle visual inspection solutions
Overview of the scenario
Based on the theory of computer vision and pattern recognition, it is self-developed by Keyi TechnologyAI - Edge Computing UnitIt fully meets the computing power support under the requirements of high definition and high speed, and has the advantages of high computing power, high stability and low power consumption, and optimizes the problem of computing power distribution during multi-camera collaborative processing.
Software and hardware platforms
1) Self-developed high-precision CCD CMOS industrial camera (camera + lens).
2) Self-developed LED light sources such as flood light source and ring light.
3) Semi-supervised learning mode.
4) Self-developed AI-edge computing unit, a high-performance embedded computing platform for industrial scenarios.
5) Build an AI cloud training platform independently.
Detect content
Bottle defects: black spots, color difference, impurities, threads, lifting rings, notches, residual materials, flashes, bubbles, holes, uneven thickness, deformation, size, inkjet code, trademark, mold number, etc.
Bottle material:p ET, PE, PP, HDPE, PC, etc.
Industry applications
Widely used:Medicines, dairy products, condiments, alcohol, beverages, daily chemicalsand other industries.
Solution Advantages:
High power
The self-developed AI-edge computing unit has faster speed, lower power consumption, stronger continuous computing ability, high temperature uninterrupted power, and stable operation.
High level of integration
It integrates optical, mechanical, electrical, computing, and software to build an AI platform with a higher degree of integration, faster computing, and stronger processing power.
Pan-data capability
The semi-supervised learning mode effectively solves the problem of small data samples and difficult labeling.
High flexibility
Supports fast switching of detection scenarios.
Easy to operate
It can be started in 3 minutes, and it can be supported by remote operation and maintenance 24/7.