Research of Convolutional Neural Networks for Text Recognition in Traffic Scene Imagery
Quanqi Liu · Advances in engineering research/Advances in Engineering Research · 2025
In the era of rapid development of intelligent transportation systems (ITS), the ability to accurately recognize text in traffic scene imagery is of utmost significance.Traffic signs, license plates, and road information in these images are crucial for traffic management, law enforcement, and navigation.For autonomous driving, it is a fundamental requirement for vehicle decisionmaking.However, previous text recognition methods, which depend on handcrafted features, struggle in complex traffic scenarios.This paper conducts an in -depth exploration of Convolutional Neural Networks (CNNs) in traffic scene text recognition.It first details the theoretical principles of CNNs, including the functions of each component in the network structure, the importance of image preprocessing, and the processes of text detection and feature recognition.Then, it showcases CNN -based research in the transportation field, particularly in traffic sign recognition.It presents the comparison of YOLOv5 and SSD models in terms of accuracy and speed, and highlights a Chinese traffic sign detection algorithm.Finally, the paper proposes future research directions such as dataset expansion and new model testing, aiming to strengthen the performance of text recognition in traffic scenes and play a role in promoting the evolution of intelligent transportation.