Text to Image Synthesis Model for Tamil Language based on Deep Learning Approach

Abstract

The history and usage of Tamil are among the longest of any Indian language. It newlinegoes back to antiquity, has its own tradition, and is steeped in a wealth of ancient literature. newlineAs the new millennium began, more than 66 million people spoke Tamil as their newlinenative tongue. In recent times applications involving image synthesis for the text information newlinehas gained attention to a greater extent. The ability to convert text into imagery newlineusing computer vision techniques involves language model and Generative Adversarial newlineArchitecture (GAN). The current state of intelligent systems research has seen them newlinetaught in universal language but has fallen short of expectations when it comes to applications newlinein regional languages. By eliminating the language barrier, regional languages newlinewill allow for a wider variety of applications to be created and additional fields of study newlineto be investigated. Existing work incorporating auto encoders had limitations in representing newlineoutput features and critical aspects of the synthesised images. In the proposed newlineresearch work a linguistic model is used to generate embedding vectors for the Tamil newlinesentence, and then a GAN is used to synthesise the representation of the image. To improve newlinethe performance of the architecture the design is optimised by considering the input newlinenoise dimensionality and L1 norm for matching features across the real images and newlinesynthesised images. The suggested architecture has several conceivable applications, newlinesuch in the fields of fashion design, photography, Computer Aideed Design (CAD), the newlineeducational sector, and more. Research concepts in native languages can be brought newlinetogether through the use of the proposed language and image synthesis deep learning newlinearchitecture. The evaluation of proposed model has achieved performance when compared newlinewith state-of-art models for image synthesis for Tamil language newline newline