The development of lane line detection is constrained by the main problems of traditional lane line detection methods, such as large computation, no visual cues and lane line occlusion. Currently, the UFast Structure-aware algorithm is proposed to effectively solve the above problems and is widely used in the field of lane line detection. While the UFast Structure-Aware algorithm has the problems that the network convolution and pooling to extract features will lose important information, and the boundary information is not sensitive enough, so the CBAM attention mechanism is added, and the above mechanism can effectively obtain richer information and extract more useful and tight features, and the L-New function pays more attention to the information of the lane shape than the Softmax function. The effectiveness of the above improvement is verified by ablation experiments without adding any computational effort. In the TuSimple benchmark dataset, the detection accuracy is improved by 0.26 percentage points compared with the original text, and the speed is comparable to the original text, making the proposed algorithm more competitive.