Abstract
There is need for a scalable and accessible smart parking system, to solve the problem of congestion, loss of business, increased air pollution, etc. This system could be used to enable parking reservation fees and demand-based parking prices to add monetary value. Camera sensors are an easy to install and cost-effective tool for collection of parking data for a smart parking system. This work presents methods for automatically locating up to 90% of the parking spots in an image and classifying the occupancy of these parking spots with a more generalizable method which does not depend on specialized datasets. These methods aim to overcome the issues of lack of generalizability and scalability in current vision based smart parking system.