This paper presents an AI-based vision system for free parking space recognition using a CCTV camera and deep learning-based object detection. The system captures video streams from a roof-mounted camera and processes them using a YOLO-based algorithm to determine parking occupancy. The detection results are transmitted to an AWS cloud server and visualized through a Progressive Web Application (PWA) for real-time monitoring. The system was deployed and tested in a large parking lot at Sungkyunkwan University, achieving over 98% accuracy in well-visible areas. The proposed system demonstrates the feasibility of low-cost, automated parking monitoring, with potential for further scalability and deployment in real-world smart parking solutions.