TSR-YOLO: A Chinese Traffic Sign Recognition Algorithm for Intelligent Vehicles in Complex Scenes.

Area of Science:

• Computer Vision
• Artificial Intelligence
• Intelligent Transportation Systems

Background:

• Traffic sign recognition is critical for autonomous driving safety.
• Environmental factors like lighting and weather degrade recognition performance.
• Existing algorithms struggle with real-world complexities.

Purpose of the Study:

• To develop a robust and efficient traffic sign detection algorithm for intelligent vehicles.
• To enhance the accuracy and reliability of traffic sign recognition under adverse conditions.
• To meet the real-time processing demands of intelligent driving systems.

Main Methods:

• Proposed a novel algorithm, TSR-YOLO, based on YOLOv4-tiny for Chinese traffic sign detection.
• Integrated an improved BECA attention mechanism and a dense SPP network into the feature extraction.
• Utilized k-means++ clustering for optimized prior box generation and added a YOLO detection layer.

Main Results:

• Achieved a detection accuracy of 96.62% on the CCTSDB2021 dataset.
• Demonstrated superior performance over the original YOLOv4-tiny, with a recall of 79.73% and F-1 Score of 87.37%.
• Maintained a high Frames Per Second (FPS) value of approximately 81, ensuring real-time capability.

Conclusions:

• The TSR-YOLO algorithm significantly improves traffic sign recognition accuracy in challenging environments.
• The proposed method effectively balances high accuracy with the real-time processing needs of intelligent vehicles.
• TSR-YOLO offers a promising solution for enhancing the safety and reliability of autonomous driving systems.