YOLO-BS: a traffic sign detection algorithm based on YOLOv8

Area of Science:

• Computer Vision
• Artificial Intelligence
• Intelligent Transportation Systems

Background:

• Traffic signs are crucial for road safety and traffic management.
• Existing detection methods struggle with accuracy and real-time performance in dynamic environments, especially with complex backgrounds and small targets.

Purpose of the Study:

• To develop an enhanced traffic sign detection algorithm (YOLO-BS) based on YOLOv8.
• To address challenges in detecting small traffic signs amidst complex backgrounds.
• To improve the real-time performance of traffic sign detection.

Main Methods:

• The study enhances the YOLOv8 framework by incorporating a small object detection layer.
• A bidirectional feature pyramid network (BiFPN) is integrated to improve multi-scale object handling.
• Experiments were conducted on the TT100K dataset to evaluate performance metrics.

Main Results:

• The proposed YOLO-BS algorithm demonstrated superior performance compared to mainstream models.
• Achieved a mean average precision (mAP50) of 90.1% and a Frames Per Second (FPS) of 78.
• The enhanced model effectively handles complex backgrounds and small detection targets.

Conclusions:

• YOLO-BS significantly improves traffic sign detection accuracy and real-time processing.
• The algorithm shows promise for enhancing safety and efficiency in intelligent transportation systems.
• Future research will focus on refining YOLO-BS for broader applications.