Text-to-Video Generation using DCGAN

Nandanwar, Aniruddha Madhukar (2024) Text-to-Video Generation using DCGAN. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (2MB) | Preview

Abstract

Generating the videos from text has actually been a growing technology and proven to important and good type of challenge for generative types of models. This study will of course explore the development and evaluation of a Deep Convolutional Generative Adversarial Network (DCGAN) for text-to-video generation, mainly focusing on floral imagery. Using the 102-category flowers dataset which consist over 7,000 annotated image-caption pairs, this study has been trained the model to convert textual descriptions into video sequences of flowers. The combination of 300-dimensional GloVe embeddings gave accurate representation of textual inputs. Training was obviously conducted on low-resolution (64x64 pixels) images due to resource constraints which is optimizing model performance with a 16 GB RAM setup. The model used and faced 438 epochs, with each epoch averaging 25.77 seconds. Results shows that the model’s capability to generate good flower videos from textual descriptions, achieving a generator loss of 1.3285 and discriminator loss of 1.2313. An interactive web application was of course been developed to showcase practical usage, enabling users to input flower descriptions and generate corresponding videos. Some types of challenges included managing computational resources and optimizing model hyperparameters for definitely good and efficient video synthesis.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Raza Abidi, Syed Muhammed UNSPECIFIED
Uncontrolled Keywords:	Text-to-Video Generation; Deep Learning; Generative Adversarial Networks (GANs); Deep Convolutional Generative Adversarial Networks (DCGANs)
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science P Language and Literature > P Philology. Linguistics > Computational linguistics. Natural language processing
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	25 Aug 2025 08:20
Last Modified:	25 Aug 2025 08:20
URI:	https://norma.ncirl.ie/id/eprint/8599

Actions (login required)

View Item