Enhancing Road Safety: YOLOv8 and Fuzzy Decision-Making for Accident Detection from Video Datasets

This study presents an innovative approach for road accident detection on highways by integrating YOLOv8, a state-of-the-art object detection model, with fuzzy decision-making mechanisms. Highway accidents pose significant challenges due to their high frequency and severe consequences, necessitating efficient and real-time detection systems. YOLOv8, known for its superior accuracy and speed in object detection, is employed to identify potential accidents from live highway surveillance footage. To enhance the robustness of the detection process, fuzzy logic is incorporated into the YOLOv8 layers. This integration aims to handle the inherent uncertainties and ambiguities in visual data, improving the model’s ability to discern accident scenarios from normal traffic conditions. The proposed system leverages the convolutional layers of YOLOv8 to extract high-level features from the input images, while the fuzzy decision-making module processes these features to evaluate the likelihood of an accident. This is achieved by defining fuzzy rules and membership functions that classify different accident indicators, such as abrupt vehicle stops, collisions, and erratic movements. The fuzzy inference system then combines these indicators to produce a confidence score for accident detection. Experimental results demonstrate that the fusion of YOLOv8 with fuzzy decision-making significantly enhances detection accuracy, reducing false positives and false negatives compared to traditional methods. This hybrid approach not only ensures real-time processing capabilities but also offers a scalable solution adaptable to various highway environments. The findings suggest that the proposed system holds substantial promise for improving traffic safety and enabling prompt emergency response through reliable accident detection.