Road damage detection based on improved YOLOv8

This study introduces an improved YOLOv8-iAD algorithm based on YOLOv8. The algorithm enhances the model's ability to fuse and extract multi-level feature information by integrating the iterative Attentional Feature Fusion (iAFF) module and the C2f module. An Adaptively Spatial Feature Fusion (ASFF) module is introduced into the detection head. The module can adjust the weights according to the relative importance of the feature graph to optimize the feature fusion process. In order to further solve the problem of inconsistent aspect ratio between targets, distance loss function (Distance-IoU, DIoU) based on the distance between center points is used to effectively improve the detection accuracy. The experimental results show that the precision and average accuracy of the improved model on the dataset RDD2022 are increased by 1.3 percentage points and 1.2 percentage points respectively.