In the advanced intelligent traffic system recognition and classification of vehicle plays a vital role. A vehicle can be recognized using various methods based on the different invariant features of the vehicle. Here vehicle classification is performed based on recognizing the unique design and pattern of vehicles tail lights. These are distinct visual features to identify specific models. Tail lights visualize a vehicle’s presence and actions, especially at night or in poor visibility. Image processing techniques like the HAAR transformation and deep learning, are used to accurately identify vehicles based on their head and tail light patterns. This system detects tail lights in real-time, providing important data for autonomous driving, traffic monitoring, and collision avoidance. The effectiveness of Haar Cascade Classifiers and deep learning models, using Feed-forward Neural Networks (FFNNs), is evaluated to improve identification accuracy and speed. As Advanced driving assistance system (ADAS) in vehicles becomes more common. By training models on diverse tail light images, an intelligent system can be developed to recognize the vehicle make and model build on tail light design in various lighting conditions. This system often identifies the unique feature like shapes and LED arrangements that help in recognition of vehicle. As compared with the performance of other machine learning models, significant improvement has achieved in our method. Experimental results clearly shows that the recognition of vehicle has improved by 2.5%–3% while using the Haar cascade feature used with yolov5.