TRAFFIC SIGN RECOGNITION USING CONVOLUTIONAL NEURAL NETWORKS

Abstract

Traffic sign recognition (TSR) plays a vital part in intelligent transport systems (ITS)
and driverless cars by allowing real-time identification and interpretation of road
signs to provide security and adherence to traffic rules. Machine learning,
traditionally applied in TSR, has been extensively applied, but ongoing developments
in deep learning, with a focus on Convolutional Neural Networks (CNNs), have
greatly enhanced accuracy and durability. CNNs are capable of learning hierarchical
feature representations automatically, decreasing the reliance on manually crafted
features. In this paper, we investigate the use of CNNs for traffic sign recognition
with a focus on model architectures, training strategies, and performance
measurement. We introduce a CNN-based TSR model that is optimized for real-time
execution with low computational cost. Our experiments show that our model
provides higher accuracy than traditional approaches. Several datasets, such as the
German Traffic Sign Recognition Benchmark (GTSRB), are utilized for validation.