Many road intersections use traffic light control systems to monitor and control the flow of automobiles. ADAS (Advanced Driver Assistance System) is a type of vehicle technology that aids drivers on the road. ADAS assists drivers and improves vehicle and road safety by reducing human error, which is common at intersections. Furthermore, recognizing the traffic light countdown timer provides the autonomous vehicle technology system with an input of the time left for the vehicle to correspondingly maneuvre and control the speed when arriving at the traffic light. As a result, the purpose of this paper is to develop a machine vision mechanism that detects and recognizes traffic lights using counter digits. By annotating and classifying all of the candidates in the 2600-frame dataset using the Cash Value Accumulation Test, the proposed method removed the background (CVAT). The files that contain the annotation information were combined with the original images before pre-processing was carried out using Roboflow, which also generates a link for importing the dataset into Google Colab for training and validation. For object detection and recognition, the YOLOv5 algorithm is used. On 100 frames per class, the method was put through its paces. The experiment produced excellent detection and recognition rates, with an average confidence rate of 80%-90% and testing dataset accuracy of 95%–100%.