Intelligent Management of Video-Monitored Parking Areas

Managing a video-monitored parking lot presents numerous challenges that must be addressed to ensure optimal functionality. These include detecting parking lines across expansive areas with multiple lots and accurately locating available spaces for drivers. The overarching objective is to streamline traffic flow to minimize drivers’ time and develop an autonomous management system that reduces the need for human oversight. Key decisions in system development involve hardware selection, such as utilizing sensors or stationary video cameras, and the choice between traditional image processing methods and modern machine learning approaches. In this study, we focus on a scenario employing a limited number of strategically positioned cameras to provide comprehensive coverage. Our approach centers on machine/deep learning techniques, employing a tailored Convolutional Neural Network solution. Additionally, we introduce an innovative dataset and implement a client-server architecture-based application on the Android platform to guide drivers to identified vacant parking spaces from their current locations.