Research On Recognition Method Of Knitted Fabric Structure Based On Deep Learning

In the knitting industry,the market share of processing batches of knitted fabric supplied by clients is huge.The type of the knitted fabric structure is an important parameter that needs to be recognized before processing the fabric,by which the knitting machine is guided.In knitted fabric structure automatic recognition,the recognition rate is influenced by uneven light,fabric rotation and yarn diameter.Developing an effective method for recognizing knitted fabric structure has great significance.A method for recognizing knitted fabric structure based on deep learning is proposed.Firstly,sample images of fabrics are captured and a knitted fabric structure image database is established.Secondly,based on transfer learning,the SqueezeNet network is trained by the database.After being trained using samples of image database,a network model of recognizing the type of knitted fabric structure is generated.The weights and biases of the SqueezeNet network are initialized before network training.In the process of network training,the weights and biases of the network are updated by the gradient descent method,during which the representative texture features of the knitted fabric are extracted.Finally,the trained network model takes fabric images as inputs and outputs the recognition results.According to the combination of both the recognition results of the fabric face and reverse side,the knitted fabric structure is recognized.The method of knitted fabric structure automatic recognition is established.The knitted fabric structure can be recognized and the recognition results can be output instantaneously according to the input images.Experiment results show that the proposed recognition method can recognize accurately the knitted fabric structure when thickness and diameter of its yarn and light are varied in fabric images.Moreover,the recognition model size is more light-weight and time-saving under the condition of high recognition rate.