Vehicle detection and identification of free slots in a garage

Abstract

The illegal or improper parking of vehicles interferes with the flow of traffic, and represents a potential hazard to both pedestrians and drivers, worldwide. Parking in unsuitable locations can also cause damage to the vehicle, so garages are the safer choice of parking for everyone. But the hard reality of the situation is that garage owners are not always skilled at efficiently handling the available parking slots for customers, due to the lack of experienced staff and drivers’ unaware of vacant spaces. Some developed countries use Internet of Things (IoT) based solutions and CCTV installations in managing parking but are very costly. Understanding these difficulties, our research endeavors to develop a cost efficient method for the detection and identification of free parking slots in garages by means of machine learning techniques. Currently, machine learning models can educate vehicle detection successfully, but they are insufficient in recognizing free parking spaces. In this study we used various algorithms such as Faster R-CNN, YOLO, and MobileNet SSD to find out which model is most suitable when we wanted to detect both vehicles and free slots accurately. After selecting the model, we expanded upon the detection process by adding on more steps to the model. Using only existing CCTV footage from the garage, our proposed algorithm does not require additional IoT devices for training and deployment, thereby significantly reducing costs for owners and providing convenience for users. On the contributing side of this research, it addresses the optimization of parking resources for various scenarios, and helps to improve both parking related services and the overall user experience. Additionally, in free space detection reliability increases to enable the establishment of parking services, which is identified as the main problem for garage users: the free space availability. Thus, a simplified management system may start to encourage private garage owners to open their facilities to the public. Finally, this research provides a useful solution to parking problems on a global scale.