Page 161 - Proceedings of the 2018 ITU Kaleidoscope

Machine learning for a 5G future

unique ID that identifies them. And within them, it will be [2] Cermele L. ―Arba detectó más de 500 mil metros
necessary to have a history of areas (square meters) cuadrados sin declarar‖. Radio Provincia - Buenos Aires,
detected as constructions over time, to later be able to Argentina. March 28, 2018
compare if any property increased in square meters or there https://www.gba.gob.ar/radioprovincia/noticias/arba_detect
are more buildings. %C3%B3_m%C3%A1s_de_500_mil_metros_cuadrados_si
n_declarar
In order to correctly perform the aforementioned tasks, it is
necessary to define an acceptable margin of error, because [3] Suarez M. ―Neuquén tiene más de un millón de
otherwise, if the difference in square meters detected was metros cuadrados sin declarar‖. Río Negro Journal.
due to an error in the detection algorithm, changes would be Neuquén, Argentina. April 26, 2017.
identified where they do not exist, that is; a false positive. https://www.rionegro.com.ar/region/neuquen-tiene-mas-de-
un-millon-de-metros-cuadrados-sin-declarar-NH2669942
Finally, those areas of 200mx200m in which a change has
been found -the ones that the surface difference is greater [4] Zhao S. ―One in four Hong Kong properties has
than the tolerance level- will be shown to the user, who illegal structures, but most owners get away with their
must determine in the software‘s final output what misdeed‖. South China Morning Post Journal. Hong Kong.
construction is new, and subsequently, verify it was January 22, 2018.
declared or not. http://www.scmp.com/news/hong-kong/law-
crime/article/2129906/one-four-hong-kong-properties-has-
6. RESULTS AND DISCUSSION illegal-structures-most

Our intention is that the software, in future implementations, [5] European Green. ―Restriction of illegal
shows the user on screen the constructions that have construction in preserved areas‖. Geneva, Italy. October 15,
changed and reframe them as a screen display, instead of 2006
just for the 200mx200m areas as it happens at this moment. https://europeangreens.eu/content/restriction-illegal-
With the current approach it is possible to limit the revision construction-preserved-areas
of the user only to those tables where there are actually
changes in buildings, however, it does not leave aside the [6] Faton Spahiu, Rineta Jashar. ―The legalization
tedious task of having to identify which construction is new process challenges on Illegal constructions in Kosovo‖.
on a map. University for Business and Technology Faculty of
Architecture and Spatial Planning. July 18, 2016.
We considered that it is possible to make some changes to https://www.researchgate.net/publication/305391468_The_l
improve this project and we propose that the corpus of egalization_process_challenges_on_Illegal_constructions_i
satellite images provided by SpaceNet could be replaced by n_Kosovo
a drone and it could take the images of the cities and send
them using the future 5G network. On the one hand, this [7] Charalabos IOANNIDIS, Christodoulos PSALTIS,
solution is more inexpensive than buying a huge dataset of Chryssy POTSIOU. ―Towards a strategy for suburban
satellite images to companies and even each city could have informal building control through automatic change
their corpus image in a database with weekly or monthly detection‖. School of Surveying Engineering, National
updates. Technical University of Athens, Greece. 2007.
https://www.researchgate.net/publication/253881637_Towa
We hope that cutting-edge technologies can be used to rds_a_strategy_for_suburban_informal_building_control_th
prevent undeclared buildings in preserved forest areas to rough_automatic_change_detection
protect the environmental biodiversity as well as be able to
avoid tragedies caused by terrible building construction. [8] Rui LIU, Yun-Fan LI. ―Research on application
mode of monitoring the illegal constructions for sale based
7. REFERENCES on multi-source data and 3S technology—a case study in
Shenzhen, China‖. Harbin Institute of Technology
[1] Sylvestre G. ―Se está notificando a 2,5 millones de Shenzhen Graduate School, Shenzhen, 518055, China.
contribuyentes para que se pongan al día‖. Gustavo Center for Assessment and Development of Real Estate,
Sylvestre Journal. Buenos Aires - Argentina. April 1 2014 Shenzhen, 518040, China. Yangze River Scientific
http://www.gustavosylvestre.com/politica/se-esta- Research Institute, Wuhan 430010, China. January 2016.
notificando-a-25-millones-de-contribuyentes-para-que-se- https://www.researchgate.net/publication/310660513_Rese
pongan-al-dia-dijo-el-titular-de-arba/ arch_on_application_mode_of_monitoring_the_illegal_con
structions_for_sale_based_on_multi-
source_data_and_3S_technology-
a_case_study_in_Shenzhen_China

– 145 –