Research And Application Of Glove Detection Technology Based On Deep Learning

The spread of the novel coronavirus has caused an international public health crisis.The demand for disposable medical supplies,including medical gloves,has increased year by year,which has brought great pressure to the production and quality inspection of gloves.Relying on the development of emerging technologies,this paper studies the glove surface defect detection method based on the object detection technology of deep learning,and proposes the Improved YOLO for the glove defect dataset.The main contents and results are as follows:(1)Aiming at the influence of the number of parameters and calculation on the efficiency of the model,a lightweight network structure YOLO-Ghost based on YOLOv5 is proposed.In the structure design of YOLO-Ghost,the Ghost convolution layer is constructed,and the C3 Ghost feature learning module is proposed to complete the reconstruction of the original network.The experimental results show that the network structure effectively reduces the overall parameters and computational complexity of the model,but it also has the problem of various defect detection accuracy decline,and the detection effect of small targets decreases greatly.(2)In order to alleviate the decline of network detection accuracy after lightweight and improve the detection effect of small targets,this paper proposes an optimization scheme of the network model.Firstly,the K-Means clustering algorithm is used to optimize the calculation method of Anchor,and the three sets of Anchor values are recalculation.The purpose is to improve the accuracy of the target box regression and accelerate the convergence of the loss function.Then,based on the CA attention mechanism and the YOLO-Ghost network structure,a feature extraction structure combined with the CA attention module is designed,which aims to improve the attention of the network to important features,and then improve the detection accuracy of the model for small targets.According to the above scheme,this paper improves the YOLO-Ghost and proposes a glove defect detection model Improved YOLO.The experimental results show that the Improved YOLO model effectively improves the detection accuracy of small targets,occupies less computing resources,and is more suitable for the deployment of industrial tasks.(3)In this paper,a glove defect detection system is built,which has five functional modules: model deployment,image data processing,video data processing,model reasoning,Fps and detection time calculation,which can meet the needs of practical applications.