<p>Traffic signal management is a complex challenge that requires advanced technology and sensor integration for optimal performance. Traditional systems, which rely on fixed signal durations, struggle to adapt to dynamic road conditions, resulting in congestion, inefficient fuel usage, and increased accident risk. This study presents an intelligent traffic signal management system employing Machine Learning (ML) and Deep Learning (DL) techniques to optimize traffic flow in response to real-time conditions. The system uses Convolutional Neural Networks (CNNs) for image processing, enabling precise vehicle detection and classification across multiple lanes. A custom algorithm adjusts signal timings dynamically based on vehicle density and type (e.g., cars, trucks, motorbikes) helps in enhancing traffic efficiency. Real-time data processing, computer vision, and sensor fusion methodologies are integrated to ensure reliable and accurate vehicle detection, particularly in complex traffic scenarios. The system is structured into three main modules: Vehicle Detection, Vehicle Classification, and Smart Traffic Light Control, each contributing to its adaptability and precision. Simulations utilizing real-world traffic data demonstrated a significant improvement in traffic flow and congestion reduction. The vehicle detection module, using VGG16 model achieved an accuracy of 98.2%, while the MobileNetV2 model and DenseNet121 model achieved 99.32% and 96.35% accuracy, respectively. Proceeding towards the vehicle classification module, using ResNet50, reached 80%. These results substantially outperformed conventional static traffic signal systems, highlighting the potential and the application of the study.</p>