Cable Lines Detection of Road Charging Piles Using Improved YOLOv8

This paper proposes an improved YOLOv8 network model to address the problem of randomly placed charging cables at roadside charging stations in recent years. In the backbone C2f module, the original convolution are replaced with SCConv (Spatial and Channel reconstruction Convolution) to form the new C2f_ScConv module, effectively reducing feature redundancy and enhancing feature representation capability. Additionally, the BiFormer attention mechanism is integrated into the head structure to focus on the subtle attributes of charging cables in complex environments, thereby enhancing network performance. Finally, the pyramid pooling module is improved by replacing SPPF with SimSPPF. Experimental results demonstrate that compared to the original YOLOv8n model, this improved algorithm shows a 9.5% improvement in mAP50 and a 4.4% improvement in mAP50-95, highlighting the effectiveness of this enhanced algorithm in accurately and quickly identifying and locating the cables of road charging piles.