Design And Implementation Of Injection Molding Product Defect Identification System Based On YOLO＿v3spp

With the development of social economy and the improvement of living standards,plastic products are more and more widely used in life.The country with the largest consumption of plastics in the world is my country,and the injection molding industry has become an important pillar industry of China’s national economy.At present,the main force in the manufacture of plastic products in China is still the traditional injection molding enterprises,but the traditional injection molding manufacturing companies are facing the disadvantages of low level of informatization and intelligence,lag in the development of production management models,and huge labor consumption.Traditional injection molding companies will urgently need With the help of artificial intelligence and other intelligent information technology means,the traditional manufacturing process and production management model will be improved and transformed,the company’s international competitiveness will be enhanced,injection molding products will be updated,and the injection manufacturing production process will be further optimized.Through interviews and investigations of plastic product manufacturers,it is found that the main difficulties faced by workers at the production site of injection molding products are as follows:First,the production status of injection molding machines,the quality of injection molding products,and the types of defects in injection molding products all rely on manual guarding and guarding.Sorting consumes a lot of manpower and has low work efficiency;secondly,the current appearance quality inspection and management system of injection molded parts has low integration and poor effect,and it is difficult to meet the production needs of modern enterprises.In view of the above problems,this paper designs and implements the quality classification system of injection molding products based on the deep learning theory and YOLO_v3 spp in combination with the production site environment of injection molding enterprises,optimizes the injection molding production process,and effectively improves the production informatization degree of injection molding enterprises.The main work of this paper is as follows:（1）Build a machine vision imaging system.Build a visual imaging system,collect the appearance images of injection molded products through line scan cameras,realize the collection of appearance images of injection molded parts,and provide support for model training and product appearance defect identification.（2）Research the YOLO_v3spp image detection algorithm and build a quality recognition network model for injection products.Conduct in-depth research on YOLO series algorithms,select YOLO_v3spp algorithm as the network for product appearance defect recognition,and make training set and test set samples to train the network model,conduct experiments on the model and analyze the experimental results.（3）Design and implement an appearance defect recognition system for injection molded parts.Combined with the production status of injection molding enterprises,analyze the functional requirements and performance requirements of the system on a feasible basis,and design the overall framework and technical architecture of the system.,user center and other functional modules for detailed design.Using SpringBoot as the development framework,combined with Nginx,MySQL,Redis and other technologies,realize various functions of the system,deploy the system and conduct functional and performance tests to verify the functionality and stability of the system.