Nowadays, the volume of highway traffic data continues to grow rapidly. To achieve efficient and accurate detection of abnormal events on highways, it is essential to establish intelligent traffic monitoring systems. However, centralizing all computational tasks in a control center results in significant bandwidth consumption and operational costs. To address this challenge, we propose a highway abnormal event detection system based on roadside edge computing. By integrating the YOLO object detection algorithm with the DeepSORT multi-object tracking algorithm, the system automates the detection of abnormal events and offloads computational tasks to edge nodes located near surveillance cameras. We present the system architecture, core technologies, and application scenarios of Antigen.