With the rapid development of autonomous driving technology, the efficient detection and accurate recognition of road surface markings have become particularly critical. To address challenges such as the diversity and complexity of road markings, as well as the susceptibility of detection accuracy to environmental factors like weather and lighting, a driving road marking detection and recognition algorithm model based on an improved YOLOv8 framework is proposed. In the backbone network, a Polynomial Kernel Introduction (PKI) module is incorporated to enable the extraction of more complex features from image data. In the neck network, a Deformable Attention (DA) mechanism is introduced to allow the model to flexibly focus on relevant parts of the image, adapting to the specific spatial structures of input data. Additionally, the original CIOU loss function is replaced with Inner-CIOU to enhance the model's generalization ability and improve detection accuracy. Experimental results demonstrate that the improved YOLOv8 network achieves a 4.4 % and 6.7 % improvement in mAP50 and mAP50-95, respectively, compared to the original network, with a detection speed of 71 Hz, verifying the effectiveness of the proposed algorithm.