基于改进YOLOv7的路面病害检测算法OA北大核心CSTPCD

In view of the substantial light and shadow variations,excessive background interference and large scale differences in highway pavement distress images,a pavement distress detection algorithm based on improved YOLOv7 is proposed.This research primarily focuses on the optimization of the ELAN module within the YOLOv7 network model,wherein an advanced amalgamation of channel and spatial attention mechanisms is employed to optimize the information extraction and elevate the network's ability in discerning and extracting important features.In order to mitigate the issue of missed detections of smaller objects in the original network model,the ACmix attention module is adeptly integrated,which amplifies the network's acuity towards smaller-scale objects.This paper introduces an innovative approach by adopting a convolutional outputs with heightened downsampling ratio to improve the precision in detecting smaller objects.The WIoUv3 is introduced to replace CIoU of the original network model to optimize the loss function,and the computation of gradient gain is constructed to attach the focusing mechanism.The experimental results show that the mean average precision(mAP)of the model based on improved YOLOv7 is increased by 4.5%relative to that of the original model,and its detection effect is better than not only the original network model but also the traditional classical object detection model.