Vehicle detection is a critical component in numerous applications, including autonomous driving, traffic monitoring, and surveillance. While traditional methods rely heavily on visible light (RGB) images, these techniques often face challenges in low-light conditions, adverse weather, and complex backgrounds. Infrared (IR) imaging offers a viable alternative, as it captures thermal signatures that are less affected by lighting and environmental factors. Unlike standard RGB images, thermal imaging has become indispensable in fields like surveillance, medicine, and industry. The ability to visualize temperature variations through thermal images has revolutionized these fields. However, in contrast to conventional RGB cameras, thermal cameras often have limited spatial resolution, which can hinder their effectiveness, and their low resolution often poses a critical bottleneck. In this context, computer vision has seen the emergence of deep learning-based approaches. Motivated by these developments, we propose a Generative Adversarial Network (GAN) based module for RGB to IR translation followed by Super-resolution module to generate high-resolution (HR) thermal images from corresponding RGB images. A GAN based module is used for domain translation, i.e., from RGB to its thermal counterpart, and a super-resolution(SR) module is used to generate its high-resolved image. In this paper, the GAN module and an SR module are integrated into a single unit that underwent end-to-end training. Although additional refinement is required, the generated outcomes represent a positive advancement. The produced thermal images adeptly depict temperature fluctuations and object outlines, demonstrating a level of similarity to real-world counterparts. This research paper aims to explore the methods of translating RGB images to IR images and then super-resolving these translated images to enhance their quality, ultimately using them for vehicle detection.