Image enhancement involves enhancing an image’s quality and visibility through different techniques like contrast adjustment, noise reduction, super-resolution, and sharpening. In unmanned aerial vehicles (UAVs), image enhancement becomes crucial due to image acquisition in challenging conditions such as low-light, haze, and fog. Specifically, while addressing low-light scenarios, the goal is to enhance the visibility and finer details of images captured in dim environments. Traditional methods often rely on handcrafted features or assumptions about the illumination model, which may lack accuracy and robustness in complex scenes. In contrast, recent advancements leverage deep learning techniques to enhance images, improving overall performance and adaptability. Our approach combines traditional and learning approaches that employ a retinex-based deep neural network. The reflectance and illumination (R&I) components are initially estimated from the image. Subsequently, the illumination component undergoes enhancement through a residual neural network, solving the vanishing gradient problem and improving image quality. Our method achieves a PSNR of 24.91 and an SSIM of 0.91, outperforming state-of-the-art methods.