Detecting traffic participants violating helmet regulations is a challenge. Based on that challenge, this article proposes a solution using deep learning models combined with computer vision to detect objects in real time from videos collected by surveillance cameras. In this article we focus on studying the Yolov8 model. YOLOv8, the latest version of the highly regarded real-time image segmentation and object detection model. Yolov8 was born to keep up with and respond to the development of deep learning and computer vision models, delivering outstanding performance in speed and accuracy. Finally, to evaluate the effectiveness of the proposed method, we collected data on videos of traffic participants, including people wearing helmets and not wearing helmets, on vnexpress.net. During the experiment, the results had a high detection accuracy of ≈80%, thereby showing that our method is effective.