General traffic administration as well as infrastructure design may benefit from real-time vehicle surveillance on motorways, roads, and streets. This study introduces Traffic Detector ,a technology that makes use of deep learning methods to automatically monitor and classify vehicles on roads using a precise and stable camera. Despite being a well-established area visual programming research, improvements in neural networks for object recognition and categorization, particularly in the last years, made this area even more intriguing owing to efficacy of these methods. It is concentrated on region-based approaches like R-CNN (Region-based Convolutional Neural Network) and regression-based methods like YOLO (You Only Look Once) and also provided each enhanced versions in the subject of motor identification is being tackled by the quickly expanding domain of supervised learning approaches. Last but not least, we have a traffic offence detection module in place that examines traffic patterns and identifies various traffic offences in real-time. The Deep Neural Network (DNN) module of OpenCV accustomed implement the complete system. With excellent accuracy, We have indeed been fortunate in locating automobiles on the roads using YOLOv4. We used a quick YOLOv4-tiny model for motorcycle riders without helmets. Real-time vehicle tracking is accomplished using Deep SORT algorithm. For vehicle detection, YOLOv4 achieves a precision of 89%, while for helmet detection, YOLOv4-tiny achieves a delicacy of 96%. YOLOv6 97%. The backbone, neck, and head are the three crucial components the majority recent iteration, known as YOLOv5.YOLOv7 is anticipated to overtake YOLO v4This review paper aims to advance sophisticated deep learning frameworks for real-time vehicle detection.