Enhancing Object Recognition in Hazy Scenes with Boundary-Constrained Dehazing and YOLOv7 Architecture

This paper presents a novel approach to improving object recognition in hazy scenes by combining boundary-constrained dehazing and YOLOv7 architecture. The current approaches encounter challenges in maintaining a trade-off between improving low-lazy images and detecting objects. In order to address this issue, the current study suggests a new method. This novel technique employs hazy images sourced from the RESIDE SOTS dataset and evaluates diverse dehazing methods based on the PSNR and SSIM metrics. The proposed method uses hazy images collected from the RESIDE SOTS dataset and compares various dehazing approaches using PSNR and SSIM metrics. Our approach enhances object recognition accuracy in hazy scenes by removing the atmospheric haze through boundary constraints and applying the state-of-the-art YOLOv7 architecture for object detection. Our experimental results demonstrate that the proposed approach outperforms other dehazing methods in terms of PSNR and SSIM metrics, achieving higher recognition accuracy for objects in hazy scenes. The proposed approach can be applied to various real-world applications such as autonomous driving, video surveillance, and environmental monitoring, where object recognition in hazy conditions is crucial.