This paper proposes a real-time vision-based blind spot detection system using sensor fusion on Raspberry Pi devices configured as edge and fog nodes. It uses YOLOv5 to spot objects and ultrasonic sensors to estimate their distance. It breaks down the terrain into a grid to understand and quantify the risky areas, assigning values that indicate how affected each part is. These are summarized with the Blind Spot Severity Index (BSSI) and Blind Spot Risk Index (BSRI). This sustaining setup is an effective alternative to traditional radar systems, assisting drivers to stay more alert in different traffic conditions. When tested, the system managed about 70% accuracy in setups with multiple objects, with a cost-effectiveness of around 180 ms, showing it can work well even on affordable hardware. The blind spot is mapped onto a grid where zero indicates clear visibility and higher values show obstruction. BSSI and BSRI quantify risk levels, enabling continuous monitoring and timely alerts to enhance driving safety.