Design And Implementation Of A Picture Generate Poems Svstem Based On Generative Adversarial Networks

With the continuous development of deep learning technology,artificial intelligence has started to enter the field of art creation.As a treasure of Chinese culture,more and more scholars have conducted research on the problem of automatic machine generation of ancient poems in recent years.Most of the current research has been on generating ancient poems by giving guide words or text sequences.Since ancient Chinese poetry often depicts a beautiful landscape,generating ancient poems by inputting images is more in line with the process of creating ancient poems.However,the task of generating ancient poems from images faces a number of challenges,including how to extract the intention of the ancient poem from the landscape image,how to generate a suitable ancient poem based on the intention of the ancient poem,and how to ensure that the generated ancient poem has a certain degree of artistry.To address these issues,the work done in this paper is as follows:This paper proposes a Rhythm-Transformer algorithmic model for the generation of Chinese poetry.By adding a visible layer to the decoder of the Transformer,the output distribution of the decoder is modified to improve the rhythm of the generated Chinese poetry.And through comparison experiments with the baseline model,the effectiveness of the ancient poem generation algorithm proposed in this paper is demonstrated under the same input.This paper proposes a picture generate Chinese poetry algorithmic model based on generative adversarial networks.By using VGG-16 network and word2vec technique,an algorithm for ancient poetry intention extraction from landscape images is implemented.A hybrid poeticassociative discriminator based on Text-CNN and hierarchical LSTM networks is proposed,which enables the Rhythm-Transformer poetry generator to generate rhythmic poems,improve the variety of poems generated and maintain the association with the input keywords.A comparison with the baseline model demonstrates the superiority of this poetry-writing algorithm.This paper designs and implements a picture generate poems system based on generative adversarial networks.The implementation process of the whole system is described in detail through requirements analysis,database design and system design.Finally,through system testing and system demonstration,it is demonstrated that the system is able to generate corresponding higher quality Chinese poetry based on the pictures uploaded by users.