Blog

Appearance Defect Detection of Bottle Body The appearance defect inspection of the bottle body is a common problem in daily production. For manufacturers in industries such as pharmaceuticals, dairy products, alcoholic beverages, seasonings, and daily chemical products, due to the increasing demand for product packaging refinement, quality, and continuous batch production, the market has put forward higher standards for bottle factory quality testing. The emergence of machine vision inspection machines based on AI algorithms, it has become a new tool to assist enterprises in efficient production.   Vision Defect Inspection Solution of Bottle Body Based on the theory of computer vision and pattern recognition, and equipped with the AI edge computing unit developed by KeyeTech, it fully meets the computing power support under the requirements of high-definition and high-speed. It has the advantages of high computing power, high stability, and low power consumption, and optimizes the problem of computing power distribution in multi camera collaborative processing.   Software & Hardware Platform (1) Self developed high-precision CCD/CMOS industrial camera (camera+lens) (2) Self developed surface light sources, circular light sources, and other LED light sources (3) Semi supervised learning mode (4) Self developed AI edge computing unit, a high-performance embedded computing platform for industrial scenarios (5) Building an AI cloud training platform independently     Detection Contents Bottle defects: black spots, color difference, impurities, threads, lifting rings, notches, residues, burrs, bubbles, holes, uneven thickness, deformation, size, spray code, trademark, mold number, etc   Bottle material: PET, PE, PP, HDPE, PC, etc   Widely Applications KeyeTech AI visual defect inspection technology is widely used in industries such as pharmaceuticals, dairy products, seasonings, alcoholic beverages, and daily chemicals. Solution Advantages High efficiency Self developed AI edge computing unit has faster speed, lower power consumption, stronger continuous computing capability, no power failure at high temperature, and stable operation.   High Integration Integrating light, machinery, electronics, computing, and software, we have built an AI platform with higher integration, faster computation, and stronger processing capabilities.   Strong data compatibility The semi supervised learning mode effectively solves the problem of small data samples and difficult labeling.   High flexibility Support fast switching of detection scenarios   Easy to operate 3 minutes to get started, with 7X24 hours of remote operation and maintenance support.      

AI algorithm - the core driving force of machine vision system Machine vision is an important branch technology in the field of artificial intelligence, and its key lies in implanting "human eyes and brain" into machines. AI algorithms provide the internal core driving force for machine vision inspection system. Regardless of the application in any scenario, machine vision needs to think, judge, and execute actions like humans. AI algorithms provide underlying data logic support for machine vision, including difficulties in target object judgment execution, complex and variable environments, feature learning and recognition, real-time performance and computing resources, and model generalization ability. With continuous increase in research and development investment and upgrading of algorithm systems, these difficulties have been gradually overcome.   Continuously Expanding Application Scenarios from Traditional to Innovation With the support of AI algorithms, the hidden properties of machine vision are released, and the application scenarios continue to expand. From the underlying logic, technology to application, AI algorithms greatly empower the intelligent application of machine vision inspection system in images, quickly establish classification models with small samples, reduce data collection and calculation costs, and improve image recognition accuracy.     At first, KeyeTech mainly applied machine vision in the field of plastic packaging defect detection. Currently, a deep collaboration among multiple industries has been formed, integrating the entire industry chain of light, machinery, electricity, computing, and software.   Through project practice in packaging containers, electronics, new energy, medicine, textiles, food and other fields, we aim to continuously expand the application scenarios of "Keye AI Machine Vision" and transform AI into true productivity.     From plastic packaging visual inspection, to visual empowerment of new energy, and to the creation of a "three in one inspection solution" for glass wine bottles, all are the fusion of "Keye machine vision". In the future, KeyeTech will also achieve comprehensive breakthroughs in AI machine vision application scenarios.

Machine vision inspection is a rapidly developing branch of artificial intelligence (AI). According to the definitions of machine vision by the Machine Vision Division of the Society of Manufacturing Engineers (SME) and the Automation Vision Division of the Robotics Industry Association (RIA), machine vision is a device that automatically receives and processes an image of a real object through optical devices and non-contact sensors to obtain the required information or to control robot motion.   Simply speaking, machine vision is using machines instead of human eyes. Machine vision simulates the eyes for image acquisition, extracts information through image recognition and processing, and finally completes the operation through the execution device.   Traditional machine vision inspection technology requires representing data as a set of features or inputting them into a prediction model to obtain prediction results. This requires completing specific actions, making it difficult to adapt to future flexible production needs, especially in scenarios where defect types are complex, subtle, and background noise is becoming increasingly difficult to apply.     After being equipped with AI deep learning function, machine vision converts the original data features into a higher-level and more abstract feature representation through multi-step feature transformation, and further inputs it into the prediction function to obtain the final result.     Machine vision based on deep learning can combine the efficiency of machine vision with the flexibility of human vision in an ideal state, thus completing detection in increasingly complex environments, especially when involving deviations or extreme environments, meeting the stringent requirements of downstream for defect accuracy and universality.  

High precision AI algorithm, meeting hundreds of subdivision scenarios Product appearance defect Inspection Mainly used for product appearance defect visual inspection system in the industrial field, including black spots, scratches, damages, foreign objects, discoloration, deformation, specifications, third phase, EAN product barcodes, etc Application scope Machine vision inspection for bottle cap bottle body, bottle preform, cup packaging, filling, spray coding (missing, blurry, not spray coded), alcohol packaging, bowl, pharmaceutical packaging; Non woven fabric, earth cloth, denim fabric, embroidered fabric; Capacitors, lighting displays, electronic components, shielding covers, connectors; COVID-19 self test kit, gloves, measuring cup, syringe, pregnancy test stick, etc. Grain quality analysis Automated, information-based, and intelligent solutions for incomplete grain analysis The KVS-G series grain quality sorting machine consists of a visual system, software system, and other module structures. When the grain enters the camera's field of view, it is photographed and the characteristics of a complete grain are comprehensively obtained through registration algorithms. Artificial intelligence algorithms are used for attribute recognition to determine whether there are problems such as disease spots, mold, sprouting, damage, and insect erosion. Application scope Quality analysis of stable agricultural products such as rice, corn, wheat, melon seeds, pine nuts, almond wood, coffee beans, etc., classified and counted according to national standards, separated and weighed (optional) for different types.   Resource recycling Based on AI algorithm classification, template matching algorithm, data preprocessing algorithm and other technologies, Keye Technology can accurately identify the materials that customers need to classify from mixed, stacked, adhesive, damaged, and dense materials, and sort them accurately. Application scope Classification of construction waste, kitchen waste, industrial waste, recycled plastics, household waste, etc.