Design And Implementation Of Painting Art Style Rendering System Based On Deep Learning

For a long time,the art of painting is mysterious to most ordinary people.Throughout the ages,countless artists have devoted their lives to paintings.Many masters have formed their own unique painting styles.With the rapid development of computer technology,especially the rapid development of Deep Learning(DL)technology in recent years,we have been able to use the Convolutional Neural Networks(CNN)to paint the styles of famous artists.Extraction and imitation,even allowing ordinary people to glimpse the mystery of artistic style,to make works with master art style,this style rendering technology is called style migration.The representation of the image in the computer is composed of one pixel,and the pixel points form a two-dimensional matrix.With the proposed convolutional neural network,different filters can be used to extract the image features,thus obtaining We want the art of painting style texture.Because the deep learning technology is too high for the general public,this paper designs and implements a WEB application based on B/S mode.Users can upload their own pictures.After selecting the style picture,the program will render the photos in the background and after the completion.Return the result to the user.The main work of this paper is as follows:1.After studying the existing style migration algorithm,after learning the network structure and characteristics of various style migration algorithms,a fast style migration algorithm is selected as the rendering module of the system background,and based on The deep learning framework TensorFlow is implemented,and finally the training results are given.2.Use Python-based Django framework to build a web application,the background uses asynchronous plus process pool model,all requests are non-blocking,support multi-user simultaneous online rendering,no need to wait in line.The front-end framework uses BootStrap+HTML5+CSS+JQuery to enhance support for mobile devices,enabling users to access and upload photos for color rendering via mobile phones,enhancing the usability of the art-style rendering system.