Vehicle identification and tracking is an extremely important function of traffic surveillance systems that is necessary for efficient traffic management and the protection of drivers and passengers. Finding and following the path of vehicles is the primary goal of this research. The goal of this research is to develop methods for the automated identification of cars in digital photographs and moving pictures. One of the numerous uses for Deep Learning, which may include fuzzy logic, neural networks, and evolutionary algorithms, is in the detection and tracking of automobiles. The purpose of this project is to apply deep learning to the problem of vehicle recognition and tracking; the primary-stage target detection techniques will be YOLOv5 and Single Shot MultiBox Detector (SSD). This is the main topic of the article. The Single Shot MultiBox Detector (SSD) model architecture is then employed as the major foundation for vehicle detection. Focus loss, in addition to the standard SSD, is an optimization component that improves feature extraction speed. Therefore, the procedure begins with a series of training procedures on the photos included inside the publicly accessible road vehicle dataset. The vehicle recognition model is then trained using YOLOv5 and SSD algorithms; these two algorithms work together to show how effective they are at detecting vehicles. Comparing the models' detection rates on different cars is the key to locating it. The fundamental objective of this study is to develop an automated technique for detecting and tracking autos in both static and dynamic scenes. In the end, the trained network model is applied to the analysis of the vehicle camera video, and the detection performance is tested experimentally. The study's results show that the approach may enhance vehicle identification success to 97.65%. From video and picture inputs, it can reliably identify vehicles.