A Smart Garage System Based On Object Detection Algorithm

When people drive back home,fetching remote control to open garage gate is a distraction,it's neither convenient nor safe.Even though automatic parking lot gate is widely used in most commercial centers and companies,which can recognize the license plate number and switch the gate accordingly,they're not applicable for civil home garage uses or installations.These gates are expensive and require customization for different use cases.With the improvement in speed and cost of computer hardware,and the breakthrough of computer vision technology,we can build a low cost and stable smart garage system for home use now.This paper discusses the algorithm and implementation of a home smart garage system,including its hardware and software designs and their implementations.The paper includes the design and tradeoff of the algorithms for license plate recognition,car front or car rear recognition.The methods and software for automatically controlling garage gate and garage light while user enters or leaves garage are discussed.This paper also addresses the methods for remote control of the garage without leaking user personal information and presents the design and implementation of a browser-server based graphical user interface for the functionalities above.In the first part,this paper introduces an object detection algorithm,and how to detect and recognize license plate number as well as car front / car rear with this algorithm,including the algorithm design and improvements.The data gathering,data cleaning,data labelling,model training procedures of this deep learning algorithm are described with details.The result shows that the presented detection algorithm could successfully identify the car plate numbers as well as car front or rear.In the second part,this paper presents the methods of hardware communication.This includes computer and embedded device communication;how to improve and control non-smart garage gate,light or alarm with Bluetooth or radio frequency signals,as well as the implementations.The experiments show that,after hardware installations,the program this paper presents could control non-smart devices with ethernet remotely and wirelessly.Following the hardware part is the graphical user interface part.This paper compares different approaches to build a user interface for a control center.It presents a browserserver based graphical user interface design,the implementation of a server-side program,and of a user interface.The server-side program this paper writes enables users to control their garage devices with smart phones,tablets or PCs.In the last part,this paper describes how to compose the algorithm,devices and user interface in a Linux operating system,to build a complete smart garage system that can be useful for home garage users.The presented system has been running for 18 months.It works stably at daytime.It has given people a lot of convenience.