國立高雄大學圖資館 |

Language: English

Back

Enhancing the Performance of VGAE Architectures for Reconstruction High-Quality 3D Faces /

Record Type:	Electronic resources : Monograph/item
Title/Author:	Enhancing the Performance of VGAE Architectures for Reconstruction High-Quality 3D Faces /Mahfoudh M Batarfi.
Author:	Batarfi, Mahfoudh M.,
Description:	1 electronic resource (103 pages)
Notes:	Source: Dissertations Abstracts International, Volume: 85-10, Section: B.
Notes:	Advisors: Mareboyana, Manohar Committee members: Ahmed, Imtiaz; Josyula, Darsana; Ramamurthy, Sreenivasan Ramasamy; Yan, Jie.
Contained By:	Dissertations Abstracts International85-10B.
Subject:	Computer science.
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=31241885
ISBN:	9798382319131

Enhancing the Performance of VGAE Architectures for Reconstruction High-Quality 3D Faces /
Batarfi, Mahfoudh M.,

Enhancing the Performance of VGAE Architectures for Reconstruction High-Quality 3D Faces /Mahfoudh M Batarfi. - 1 electronic resource (103 pages)

Source: Dissertations Abstracts International, Volume: 85-10, Section: B.

This research investigates methods to enhance the performance of Variational Graph Autoencoder (VGAE) architectures for reconstructing high-quality 3D face data. Accurate reconstruction of 3D facial data is crucial for various applications such as facial recognition, virtual reality, and computer graphics. However, existing methods often struggle with limitations in dataset quality and handling large-scale datasets efficiently.This research addresses these challenges through three key contributions. First, it develops a comprehensive methodology to generate a diverse and realistic 3D face dataset that accurately captures facial expressions and variations. This dataset is specifically optimized for effective training of VGAE architectures. Second, the research focuses on adapting VGAE architectures with Graph Convolutional Network (GCN) layers. This integration allows VGAEs to efficiently handle large-scale 3D face datasets while maintaining computational efficiency through memory optimization techniques. Finally, the research evaluates the effectiveness of different GCN layer types within the VGAE framework. It analyzes the impact of these layer types on reconstruction accuracy, measured by minimizing validation loss and computational efficiency, considering execution time within a single GPU environment.The findings of this research provide valuable insights into the effectiveness of different GCN layer types for improving the reconstruction performance of VGAEs in 3D face reconstruction tasks. This work advances 3D face reconstruction methodologies by establishing a solid foundation for future research on applying VGAE architectures with GCN layers.

English

ISBN: 9798382319131Subjects--Topical Terms:

199325
Computer science.
Subjects--Index Terms:

Fine-tuning hyperparameters

Enhancing the Performance of VGAE Architectures for Reconstruction High-Quality 3D Faces /
LDR:03196nmm a22004093i 4500 001 676313
005 20250420214341.5
006 m o d
007 cr|nu||||||||
008 250901s2024 miu||||||m |||||||eng d
020 $a 9798382319131
035 $a (MiAaPQD)AAI31241885
035 $a AAI31241885
040 $a MiAaPQD $b eng $c MiAaPQD $e rda
100 1 $a Batarfi, Mahfoudh M., $e author. $3 990079
245 1 0 $a Enhancing the Performance of VGAE Architectures for Reconstruction High-Quality 3D Faces / $c Mahfoudh M Batarfi.
264 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2024
300 $a 1 electronic resource (103 pages)
336 $a text $b txt $2 rdacontent
337 $a computer $b c $2 rdamedia
338 $a online resource $b cr $2 rdacarrier
500 $a Source: Dissertations Abstracts International, Volume: 85-10, Section: B.
500 $a Advisors: Mareboyana, Manohar Committee members: Ahmed, Imtiaz; Josyula, Darsana; Ramamurthy, Sreenivasan Ramasamy; Yan, Jie.
502 $b D.Sc. $c Bowie State University $d 2024.
520 $a This research investigates methods to enhance the performance of Variational Graph Autoencoder (VGAE) architectures for reconstructing high-quality 3D face data. Accurate reconstruction of 3D facial data is crucial for various applications such as facial recognition, virtual reality, and computer graphics. However, existing methods often struggle with limitations in dataset quality and handling large-scale datasets efficiently.This research addresses these challenges through three key contributions. First, it develops a comprehensive methodology to generate a diverse and realistic 3D face dataset that accurately captures facial expressions and variations. This dataset is specifically optimized for effective training of VGAE architectures. Second, the research focuses on adapting VGAE architectures with Graph Convolutional Network (GCN) layers. This integration allows VGAEs to efficiently handle large-scale 3D face datasets while maintaining computational efficiency through memory optimization techniques. Finally, the research evaluates the effectiveness of different GCN layer types within the VGAE framework. It analyzes the impact of these layer types on reconstruction accuracy, measured by minimizing validation loss and computational efficiency, considering execution time within a single GPU environment.The findings of this research provide valuable insights into the effectiveness of different GCN layer types for improving the reconstruction performance of VGAEs in 3D face reconstruction tasks. This work advances 3D face reconstruction methodologies by establishing a solid foundation for future research on applying VGAE architectures with GCN layers.
546 $a English
590 $a School code: 1254
650 4 $a Computer science. $3 199325
650 4 $a Computer engineering. $3 212944
653 $a Fine-tuning hyperparameters
653 $a Graph Convolutional Network
653 $a Graph Neural Networks
653 $a Optimization hyperparameters
690 $a 0984
690 $a 0464
710 2 $a Bowie State University. $b Computer Science. $e degree granting institution. $3 990080
720 1 $a Mareboyana, Manohar $e degree supervisor.
773 0 $t Dissertations Abstracts International $g 85-10B.
790 $a 1254
791 $a D.Sc.
792 $a 2024
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=31241885