Comparative Analysis of the Generative Adversarial Network Model for Image Synthesis for Video Processing
Journal of Sensors, IoT & Health Sciences · 2025
DOI: 10.69996/jsihs.2025005 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Sensors, IoT & Health Sciences, 2025. Verfasst von Biswanath Saha.
Abstract (Originalsprache)
Generative Adversarial Networks (GANs) have emerged as a groundbreaking tool in the field of machine learning, primarily due to their impressive capability to generate high-quality, realistic images. This paper explores the capabilities of GANs for image synthesis and examines their applications in various domains beyond image generation, including video synthesis, data augmentation, and even drug discovery. GANs consist of two neural networks— a generator and a discriminator— which are trained simultaneously in a game-theoretic framework to improve the quality of the generated outputs. The generator creates synthetic images, while the discriminator This paper investigates the performance of different Generative Adversarial Networks (GANs) for image synthesis and classification tasks, with a focus on Weighted GAN. We compare Weighted GAN with other GAN variants, including Standard GAN, WGAN, and cGAN, across multiple publicly available datasets such as CIFAR-10, MNIST, Fashion MNIST, CelebA, ImageNet, and LSUN. The evaluation metrics include FID score, Inception score, classification accuracy, precision, recall, F1-score, training time, and cross-entropy loss. Our results show that Weighted GAN consistently outperforms other models, achieving superior image quality, faster convergence, and higher classification performance. The Weighted GAN also exhibits low mode collapse, making it a robust choice for generating diverse and realistic images. This study highlights the efficiency and effectiveness of Weighted GAN for both image synthesis and classification tasks, offering a promising approach for a range of computer vision applications
Zitieren
APA
Saha, B. (2025). Comparative Analysis of the Generative Adversarial Network Model for Image Synthesis for Video Processing. Journal of Sensors, IoT & Health Sciences. https://doi.org/10.69996/jsihs.2025005
Vancouver
Saha B. Comparative Analysis of the Generative Adversarial Network Model for Image Synthesis for Video Processing. Journal of Sensors, IoT & Health Sciences. 2025. doi:10.69996/jsihs.2025005
BibTeX
@article{Saha2025,
author = {Saha, Biswanath},
title = {Comparative Analysis of the Generative Adversarial Network Model for Image Synthesis for Video Processing},
journal = {Journal of Sensors, IoT & Health Sciences},
year = {2025},
doi = {10.69996/jsihs.2025005}
}
EUVIMED