A Novel Lightweight Real-Time Traffic Sign Detection Integration Framework Based on YOLOv4

Area of Science:

• Intelligent Transportation Systems
• Computer Vision
• Deep Learning

Background:

• Traffic sign detection is crucial for intelligent transportation but faces challenges in feature extraction, method selection, and real-world application.
• Existing methods struggle with scale and illumination variations, limiting deployment in realistic scenarios.

Purpose of the Study:

• To propose a lightweight, real-time traffic sign detection framework using deep learning (YOLO).
• To optimize the framework for reduced computational overhead and enhanced information transfer.
• To improve detection efficiency, generalization, robustness, and performance under varying environmental conditions.

Main Methods:

• Developed a lightweight integration framework based on YOLO (You Only Look Once).
• Optimized network computational overhead to reduce latency.
• Facilitated information transfer and sharing across diverse network levels.

Main Results:

• The proposed model demonstrated improved detection efficiency and robustness.
• Achieved enhanced performance in challenging environments with scale and illumination variations.
• Successfully balanced detection performance with deployment difficulty, reducing computational cost.

Conclusions:

• The framework offers a viable solution for real-time traffic sign detection on edge devices with limited hardware.
• The research provides a practical approach for deploying advanced AI in autonomous driving and intelligent transportation.
• The model shows potential for application in artificial intelligence and autonomous driving industries.