With a market size of 100 million people, autonomous vehicles are poised to become a popular solution in the world in general and in Vietnam in particular shortly. However, the self-driving car system faces a significant challenge due to the limited availability of data on the current traffic conditions in Vietnam. Specifically, datasets and models for sign recognition in Vietnam are scarce, potentially impacting the widespread adoption of autonomous vehicles in the country. This article addresses the detection of common road traffic sign patterns using two models: Yolov8 and Faster R-CNN. The dataset used was collected by us in three major cities in Vietnam and includes four main types of signs: Prohibited Signs, Warning Signs, Regulatory Signs, and Guide Signs, totaling 23 classes. Both the Yolov8 and Faster R-CNN models demonstrated strong performance with an average accuracy (mAP@50) of 95.9% and 95.3%, respectively.