Study on a method for detecting and tracking multiple traffic signals at the same time using YOLOv7 and SORT object tracking
Quan Pham Hong, Thien Nguyen Luong, Tung Pham Xuan, Manh Tran Duc, Ngoc Pham Van Bach, Tuan Pham Minh, Tuyen Bui Trong, Hoang Le Huy · 2023
Improving transportation safety and efficiency is currently a top priority worldwide. With the increasing use of autonomous vehicles, it is crucial to develop real-time traffic sign monitoring systems to assist self-driving cars in recognizing and making safe decisions on the road. However, detecting and tracking multiple traffic signs poses challenges due to variations in numbers, sizes, and colors, especially when there are similar warning contents. Therefore, the development of a real-time traffic sign monitoring system using advanced image recognition technology is highly significant. This study introduces the utilization of the YOLOv7 model combined with SORT object tracking to simultaneously track multiple traffic signals. The focus is on a system that provides speed limits for self-driving cars based on different types of signs, such as R.120 and P.127, placed closely together. The YOLOv7 model will be used for sign recognition, while SORT object tracking assigns individual IDs to track the signs. Based on each specific case, the maximum speed for self-driving cars will be determined. This method will greatly benefit self-driving cars in detecting multiple traffic signs in real-time.