Efficient Parking Lot Management System and Aberrant Behavior Detection Only Using Surveillance Cameras

This paper analyzes surveillance camera images already installed in parking lots without building additional infrastructure to analyze the location of vehicles in parking lots and explain an effective method of detecting abnormal behavior of vehicles. In this work, we proposed a method which analyzes the image using a deep learning model and expresses the location of the vehicle as integrated grid map coordinate. Additionally introduce a method for detecting abnormal behavior based on the generated grid map information and image analysis results. We conducted some experiment using four FHD videos, as a result, we proposed a lightweight system that performs in real time while increasing the accuracy of location estimation and abnormal behavior detection.