Real-Time Detection and Classification of Street Signs Using Deep Learning with Voice Alert system

Road accidents caused by human factors are a major concern for road safety, and tired or drowsy drivers can often misinterpret road signs, leading to dangerous situations. To address this issue, street sign recognition systems can provide drivers with real-time information about the road ahead, enhancing safety on the road. Convolutional neural networks (CNNs) have proven to be effective for street sign recognition, but different types of street signs may require different models to extract their unique features. Unfortunately, more complex models may be too time-consuming to meet real-time requirements. This paper proposes a street sign recognition system based on CNNs, specifically the LeNet model, which efficiently classifies German street signs into the correct category. Additionally, the system converts the text of the classified sign into speech, providing drivers with audible information about upcoming signposts. By leveraging the strengths of the LeNet model, our system provides accurate and efficient street sign recognition while meeting real-time requirements.