In view of the high number of parameters and large amount of computation in the existing models, it is difficult to embed the Conical bucket detection model into edge devices. A lightweight YOLOv8n-CTM(Conical Traffic Marker) model is proposed. Firstly, C2F-AKconV is constructed by introducing Alterable kernael Convolution (AKConv), and all c2f modules are replaced in the backbone network to enhance the feature extraction ability of the model and reduce the number of parameters and frame rate of the model. Secondly, the weighted bidirectional pyramid network (BiFPN) is introduced to replace the Neck part of the original model. BiFPN significantly enhances the model's performance in the object detection task by optimizing the multi-scale feature fusion mechanism and detecting the features of the input images. Subsequently, the LSCD(lightwerght scalable shared convolutional detection head) lightweight detection head was designed based on shared convolution and adaptive feature scaling, which enabled the model to reduce the consumption of computational load and parameter quantity while maintaining the detection accuracy. Finally, the intelligent intersection and union ratio loss function Wise-IoU v3 is introduced to improve the training accuracy. The parameter quantity and frame rate of YOLOv8N-CTM decreased by 70.6% and 47.5% respectively compared with the original model YOLOv8n, while the mAP0.50 increased by 0.9%. YOLOv8n-CTM has significant advantages in the detection of conical traffic signs and can provide certain technical support for urban traffic management and autonomous driving of vehicle safety.