Welcome to the InfoSci Platform
Text to Image Synthesis Using Multistage Stack GAN

Text to Image Synthesis Using Multistage Stack GAN

V. Dinesh Reddy, Yasaswini Desu, Medarametla Sindhu, Chilukuri Vamsee, Neelissetti Girish
Copyright: © 2023 |Pages: 16
ISBN13: 9781668469378|ISBN10: 1668469375|ISBN13 Softcover: 9781668469385|EISBN13: 9781668469392
DOI: 10.4018/978-1-6684-6937-8.ch010
Cite Chapter Cite Chapter

MLA

Dinesh Reddy, V., et al. "Text to Image Synthesis Using Multistage Stack GAN." Handbook of Research on AI Methods and Applications in Computer Engineering, edited by Sanaa Kaddoura, IGI Global, 2023, pp. 206-221. https://doi.org/10.4018/978-1-6684-6937-8.ch010

APA

Dinesh Reddy, V., Desu, Y., Sindhu, M., Vamsee, C., & Girish, N. (2023). Text to Image Synthesis Using Multistage Stack GAN. In S. Kaddoura (Ed.), Handbook of Research on AI Methods and Applications in Computer Engineering (pp. 206-221). IGI Global. https://doi.org/10.4018/978-1-6684-6937-8.ch010

Chicago

Dinesh Reddy, V., et al. "Text to Image Synthesis Using Multistage Stack GAN." In Handbook of Research on AI Methods and Applications in Computer Engineering, edited by Sanaa Kaddoura, 206-221. Hershey, PA: IGI Global, 2023. https://doi.org/10.4018/978-1-6684-6937-8.ch010

Export Reference

Mendeley
Favorite

Abstract

Many recent studies on text-to-image synthesis decipher approximately 50% of the problem only. They failed to compute all the imperative details in it. This chapter presents a solution using stacked generative adversarial networks (GAN) to generate lifelike images based on the given text. The stage-I GAN creates a distorted images by depicting the rudimentary/basic colours and shape of a scene predicted on text illustration. Stage-II GAN ends up on generating high-resolution images with naturalistic features using Stage-I findings and the text description as inputs. The output generated by this technique is more credible than many other techniques which are already in use. More importantly, stack GAN produces 256 x 256 images based on the text descriptions, while the existing algorithms produces 128 x 128.

Request Access

You do not own this content. Please login to recommend this title to your institution's librarian or purchase it from the IGI Global bookstore.