LVRNet: Lightweight Image Restoration for Aerial Images under Low Visibility (Student Abstract)

Learning to recover clear images from having a combination of degrading factors is challenging task. That being said, autonomous surveillance in low visibility conditions caused by high pollution/smoke, poor air quality index, light, atmospheric scattering, and haze during blizzard, etc, becomes even more important prevent accidents. It thus crucial form solution that can not only result high-quality image but also which efficient enough be deployed for everyday use. However, the lack proper datasets available tackle this task limits performance previous methods proposed. To end, we generate LowVis-AFO dataset, containing 3647 paired dark-hazy images. We introduce new lightweight deep learning model called Low-Visibility Restoration Network (LVRNet). outperforms restoration with latency, achieving PSNR value 25.744 an SSIM 0.905, hence making our approach scalable ready practical