In recent years, the rapid increase in vehicular traffic has heightened the urgency for intelligent accident detection systems capable of real-time operation. The timely identification of road accidents and immediate response mechanisms are vital for minimizing casualties, reducing secondary accidents, and enhancing overall road safety. Artificial Intelligence (AI), particularly deep learning-based object detection models, has emerged as a powerful tool in this domain. Among these, the YOLO (You Only Look Once) family of algorithms has gained significant attention for its speed and accuracy. The latest iteration, YOLOv11, incorporates several architectural enhancements and optimizations that make it highly effective for real-time video analytics in complex traffic environments. This study proposes an accident detection system built upon the YOLOv11 framework to identify critical road entities such as vehicles, pedestrians, and traffic-related objects. By processing real-time video feeds, the system continuously monitors traffic patterns to detect anomalies indicative of accidents—such as abrupt halts, collisions, or erratic vehicle trajectories. The high-speed inference capability of YOLOv11 allows for near-instantaneous detection, which is essential for triggering automated alerts to emergency responders and traffic management authorities without delay. Experimental evaluations of the proposed system demonstrate promising results, achieving a precision of 90%, recall of 88%, and an F1-score of 86% at a confidence threshold of 0.7. These metrics indicate the robust performance of the system even in complex, dynamic traffic scenarios involving occlusions, dense vehicle movement, and varying environmental conditions. The results show that the integration of YOLOv11 significantly enhanced detection reliability while maintaining real-time processing efficiency.