This paper is related for ADAS (Advanced Driving Assistance System) when a driver recognizes the parking situations. A wide-angle camera is mounted on the front, rear, left and right sides of the vehicle, and an image converted to the top view is used based on the information obtained by calibrating the camera. Considering that the visible range of the AVM system varies depending on the mounting position of the vehicle, the actual distance of the AVM image is measured and designated as ROI (the Region of Interest) to the front and rear, right and left 2 meters from the vehicle. (3 areas in the horizontal direction and 3 areas in the vertical direction), determines whether the parking line is crossed for each area by using deep learning, and finally a warning signal generated for the result.