MACGAN: An All-in-One Image Restoration Under Adverse Conditions Using Multidomain Attention-Based Conditional GAN

Abstract

Various vision-based tasks suffer from inaccurate navigation and poor performance due to inevitable problems, such as adverse weather conditions like haze, fog, rain, snow, and clouds affecting ground and aerial navigation, as well as underwater images being degraded with blue-green tones and mud affecting marine navigation. While most techniques in the literature are designed to restore only one or a few degradations using a single model, using separate architectures for each restoration process is computationally expensive. To address this, an all-in-one Multidomain Attention-based Conditional Generative Adversarial Network (MACGAN) is proposed to improve scene visibility for optimal ground, aerial, and marine navigation, using the same set of parameters across all domains. The MACGAN is a lightweight network with four encoder and decoder blocks and multiple attention blocks in between, which enhances the image restoration process by focusing on the most important features. The performance of MACGAN is compared qualitatively and quantitatively with various state-of-the-art image-to-image translation models, all-in-one adverse weather removal models, and single-effect removal models. Further, MACGAN is tested on real-world unseen image domains, such as smog, dust, fog, rain, snow, and lightning. The results demonstrate that MACGAN can improve scene visibility and has good generalizability. An ablation study is also conducted to assess the impact of the discriminator and attention blocks in MACGAN, and the results confirm the effectiveness of the proposed architecture, which includes a discriminator and three attention blocks.