This study presents an innovative tool designed to automate the detection of yellow lights and analyze vehicle behavior at intersections, leveraging advanced deep learning and computer vision techniques. Using traffic footage collected from intersections in Utah and Florida, the tool processes multi-camera video feeds to evaluate driver responses during the yellow light phase. The integration of YOLOv8, a state-of-the-art object detection model, enables accurate identification of vehicles in real-time, while OpenCV facilitates video frame extraction and synchronization across multiple cameras. This ensures comprehensive traffic analysis, even in complex environments. The tool optimizes data storage and processing efficiency by automating the identification of yellow light occurrences and discarding video frames without vehicle activity. Behavioral analysis focuses on determining whether vehicles decelerate, stop, or accelerate during the yellow light phase, with virtual distance markers used to estimate speed and acceleration. These insights provide valuable data for modifying traffic signal timings to align with real-world driver behavior. Preliminary results demonstrate the tool's potential to enhance traffic flow and safety at intersections. The findings indicate that adaptive adjustments to yellow light durations, informed by observed driver behavior, could reduce congestion and the risk of accidents. Future applications include integrating real-time adaptive traffic signal systems that adjust signal timings dynamically based on live traffic data. This research also lays the groundwork for further incorporating vehicle-to-infrastructure communication to optimize urban traffic management. By combining machine learning and computer vision, this study introduces a scalable, efficient, and accurate solution for addressing the challenges of modern traffic systems. It establishes a strong foundation for future research in intelligent transportation systems, aiming to improve traffic safety and efficiency through data-driven decision-making.