Automated driving is a key topic of concern for the whole society, and road sign recognition technology is indispensable for automated driving systems. Some researchers have found that some modifications of road signs may cause errors in road sign identification, so the topic of this paper is to use the attention strategy to improve the performance of road sign recognition. This paper uses the GTSRB-German Traffic Sign Recognition Benchmark from Kaggle as the dataset. The dataset contains 43 classes with a total of more than 50,000 images of 30*30 pixels. The method adopts the convolutional neural network which is built based on VGG-16 and uses SENet as an attention generative network. Without the attention strategy, the highest accuracy obtained was 98.66% with 45 epochs. By adding the attention network, the result is 98.9%. The result of the study shows that SENet has better performance because it improves performance not only in the spatial dimension but also in the relationship between feature channels.