A Deep Dive into Generative Adversarial Networks designed for Dehazing: An In-Depth Survey and Eight Efficient Pix2pix Models

In recent years, there has been a growing trend in the automated dehazing of images, owing to the real-world significance of advanced technologies. In light of the current relevance of this field, our work aims to provide a comprehensive survey of Generative Adversarial Network (GAN)-based dehazing networks developed to date. The survey focuses on the effectiveness of GANs in computer vision applications and seeks to present the current state-of-the-art to researchers in this domain, shedding light on unexplored areas that require further improvement. In addition to this survey, here we have introduced eight lightweight advanced Pix2pix models that strike a balance between computational efficiency and network performance. We also conducted in-depth qualitative and quantitative analyses of the results obtained from these models, utilizing various benchmarking databases. We employed both no-reference and full-reference quantitative parameters to validate the effectiveness of these models in addressing the dehazing problem.