An Algorithm for Street Parking Sign Rule Generation

Abstract

Autonomous driving and autonomous parking have been actively investigated for years. However, for autonomous street parking, the questions of how to determine where to park and how long can park have never been addressed. To solve these problems, we design a pipeline of detecting the valid street parking signs, recognizing the plain text on the signs, and understanding their semantic meanings. Specifically, in this thesis, we propose a street parking sign rule generation engine, which can figure out the allowed parking time and payment information for each vehicle category. To the best of our knowledge, this is the first work on designing the algorithm for street parking rule generation from street parking sign pictures. By utilizing the proposed work, our smart street parking system can detect and recognize the texts on street parking signs and extract street parking rules. The preliminary results illustrate that our system can successfully generate accurate street parking rules from different types of street parking signs.