Research On Flower Image Generation Method Based On Generative Adversarial Networks

Nowadays,many models of machine learning and deep learning have been widely used in our production and life.Images are an important carrier of information dissemination,and people can obtain a lot of external information through images.However,the early design of some agricultural production practices requires the use of flower images with low quality requirements to guide the production practice,thus avoiding the consumption of a large amount of manpower,material resources and financial resources due to the late reaction.Therefore,this paper hopes to use the generated image to generate different colors of floral images to be applied to the design of garden flower landscape renderings and campus art panels,to help solve the problem of how to place garden flowers and how to design campus art panels.Thereby greatly reducing the consumption of time,manpower,material resources and financial resources.This paper takes flowers as the research object.The research work is mainly carried out from the following aspects:First,read a large number of references to understand the basic structure of the generated confrontation network and the types of different generation-oriented networks.Then,the parameters of the generated confrontation network are set,and the generated confrontation network is trained.The quality of flower images generated by different generation-oriented network models is analyzed from a quantitative perspective.It is better to compare the quality of different generated images against the network-generated images and to obtain the flower images generated by DCGAN.Finally,the DCGAN network model is used to generate different color flower images,and the generated flower images are used to design unique garden flower landscape renderings and campus art panels renderings,which are used to guide the real flower placement and reality.The display panels are drawn to make our lives rich and colorful.