Unveiling Facial Fidelity: A Novel Approach to Synthesizing High-Quality Face Images Using Generative Adversarial Networks

In the field of computer vision, generating realistic images from sketches has garnered considerable attention due to its wide-ranging applications in art, design, and facial recognition systems. With increasing demand for more sophisticated image synthesis, advanced methodologies are essential to b...

Full description

Saved in:
Bibliographic Details
Main Authors: Modafar Ati, Muhammad Ahmed Hassan, Muhammad Usman Ghani
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/11030461/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850217745433493504
author Modafar Ati
Muhammad Ahmed Hassan
Muhammad Usman Ghani
author_facet Modafar Ati
Muhammad Ahmed Hassan
Muhammad Usman Ghani
author_sort Modafar Ati
collection DOAJ
description In the field of computer vision, generating realistic images from sketches has garnered considerable attention due to its wide-ranging applications in art, design, and facial recognition systems. With increasing demand for more sophisticated image synthesis, advanced methodologies are essential to bridge the gap between sketches and high-fidelity images. This study proposes a novel methodology that comprises two key modules: the Generator and the Discriminator. The Generator is designed to transform sketches into images, while the Discriminator extracts 512-dimensional face encodings from both generated and ground truth images. Leveraging pre-trained ResNet-101, the Discriminator uses Mean Squared Error (MSE) loss to compare the face encodings, reinforcing the Generator to produce images with face encodings closely resembling the ground truth. The research utilizes two datasets: Labeled Faces in the Wild (LFW) for training the ResNet-101 CNN model for face recognition, and CelebFaces Attributes Dataset (Celeb-A) for training the sketch-to-image generator. To assess the quality of generated images, Signal-to-Noise Ratio (SNR) and Peak Signal-to-Noise Ratio (PSNR) evaluation measures are used. The SNR steadily increases over the epochs, indicating enhanced performance, and reaches a peak value of 0.334. This indicates the model’s efficacy in reducing noise and enhancing signal strength, leading to improved overall performance. The highest achieved PSNR value of 0.334 signifies an enhancement in the quality of the reconstructed signal. In particular, the proposed model exceeds previous models, with accuracies of 99.9% on the LFW dataset and 99.4% on the Celeb-A dataset, respectively.
format Article
id doaj-art-8bc64df2b8c8422bb7ff344a4b55a8aa
institution OA Journals
issn 2169-3536
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-8bc64df2b8c8422bb7ff344a4b55a8aa2025-08-20T02:07:59ZengIEEEIEEE Access2169-35362025-01-011310348210349610.1109/ACCESS.2025.357894011030461Unveiling Facial Fidelity: A Novel Approach to Synthesizing High-Quality Face Images Using Generative Adversarial NetworksModafar Ati0https://orcid.org/0000-0001-9250-6225Muhammad Ahmed Hassan1Muhammad Usman Ghani2https://orcid.org/0000-0001-6733-2569Computer Science and Information Technology, College of Engineering, Abu Dhabi University, Abu Dhabi, United Arab EmiratesAlkhwarizmi Institute of Computer Science, University of Engineering and Technology, Lahore, Punjab, PakistanAlkhwarizmi Institute of Computer Science, University of Engineering and Technology, Lahore, Punjab, PakistanIn the field of computer vision, generating realistic images from sketches has garnered considerable attention due to its wide-ranging applications in art, design, and facial recognition systems. With increasing demand for more sophisticated image synthesis, advanced methodologies are essential to bridge the gap between sketches and high-fidelity images. This study proposes a novel methodology that comprises two key modules: the Generator and the Discriminator. The Generator is designed to transform sketches into images, while the Discriminator extracts 512-dimensional face encodings from both generated and ground truth images. Leveraging pre-trained ResNet-101, the Discriminator uses Mean Squared Error (MSE) loss to compare the face encodings, reinforcing the Generator to produce images with face encodings closely resembling the ground truth. The research utilizes two datasets: Labeled Faces in the Wild (LFW) for training the ResNet-101 CNN model for face recognition, and CelebFaces Attributes Dataset (Celeb-A) for training the sketch-to-image generator. To assess the quality of generated images, Signal-to-Noise Ratio (SNR) and Peak Signal-to-Noise Ratio (PSNR) evaluation measures are used. The SNR steadily increases over the epochs, indicating enhanced performance, and reaches a peak value of 0.334. This indicates the model’s efficacy in reducing noise and enhancing signal strength, leading to improved overall performance. The highest achieved PSNR value of 0.334 signifies an enhancement in the quality of the reconstructed signal. In particular, the proposed model exceeds previous models, with accuracies of 99.9% on the LFW dataset and 99.4% on the Celeb-A dataset, respectively.https://ieeexplore.ieee.org/document/11030461/Computer visiondeep learninggenerative adversarial networkimage generationimage processing
spellingShingle Modafar Ati
Muhammad Ahmed Hassan
Muhammad Usman Ghani
Unveiling Facial Fidelity: A Novel Approach to Synthesizing High-Quality Face Images Using Generative Adversarial Networks
IEEE Access
Computer vision
deep learning
generative adversarial network
image generation
image processing
title Unveiling Facial Fidelity: A Novel Approach to Synthesizing High-Quality Face Images Using Generative Adversarial Networks
title_full Unveiling Facial Fidelity: A Novel Approach to Synthesizing High-Quality Face Images Using Generative Adversarial Networks
title_fullStr Unveiling Facial Fidelity: A Novel Approach to Synthesizing High-Quality Face Images Using Generative Adversarial Networks
title_full_unstemmed Unveiling Facial Fidelity: A Novel Approach to Synthesizing High-Quality Face Images Using Generative Adversarial Networks
title_short Unveiling Facial Fidelity: A Novel Approach to Synthesizing High-Quality Face Images Using Generative Adversarial Networks
title_sort unveiling facial fidelity a novel approach to synthesizing high quality face images using generative adversarial networks
topic Computer vision
deep learning
generative adversarial network
image generation
image processing
url https://ieeexplore.ieee.org/document/11030461/
work_keys_str_mv AT modafarati unveilingfacialfidelityanovelapproachtosynthesizinghighqualityfaceimagesusinggenerativeadversarialnetworks
AT muhammadahmedhassan unveilingfacialfidelityanovelapproachtosynthesizinghighqualityfaceimagesusinggenerativeadversarialnetworks
AT muhammadusmanghani unveilingfacialfidelityanovelapproachtosynthesizinghighqualityfaceimagesusinggenerativeadversarialnetworks