Automatic Detection of Road Cracks using EfficientNet with Residual U-Net-based Segmentation and YOLOv5-based Detection

The main factor affecting road performance is pavement damage. One of the difficulties in maintaining roads cracking. Credible and reliable inspection heritage structural health relies heavily on crack detection surfaces. To achieve intelligent operation maintenance, essential to traffic safety. cracks using computer vision has gained popularity recent years. Recent technological breakthroughs general deep learning algorithms have resulted improved results discipline detection. In this paper, two techniques for object identification segmentation are proposed. EfficientNet with residual U-Net technique suggested segmentation, while YOLO v5 algorithm offered correctly separate cracks, a network used. Road accuracy were enhanced by optimising model's hyperparameters increasing feature extraction structure. algorithm's compared state-of-the-art algorithms. work achieves 99.35% accuracy.