An essential part of intelligent transportation systems and traffic management is lane-based vehicle recognition. This paper introduces a real-time vehicle detection and classification system that uses deep learning models. Results from experiments show that the system can manage a variety of traffic situations, such as occlusions, high vehicle densities, and changing illumination conditions. The accuracy of traffic analysis is greatly increased by this lane-based detection system, which also supports real-time decision-making in traffic monitoring and offers insightful information for smart city applications.