Detecting advertising on building fa ades with computer vision

Outdoor advertising influences the visual environment of any modem city. Advertising and information signs on building facades are one of the types of outdoor advertising. As a rule, there are laws and design codes in cities that define permissible look of such signs. At the same time, in metropolises, there is a problem of timely detection of advertising constructions on facades that violate these rules. City-scale monitoring of facade conditions is beyond the capabilities of any city's authorities. To address this issue, we propose a solution which combines street-view maps and machine learning to automate the process of searching for law-breaking advertising objects on building facades. We develop a dataset for a machine learning model and a set of checks for detected advertising objects to check their legality. The resulting approach can provide data for future research and help maintain coherent urban visual environment. (C) 2019 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC -ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under responsibility of the scientific committee of the 8th International Young Scientist Conference on Computational Science.