Naslov
Detection of strong mine presence indicators using intelligent algorithms

Autori
Horvat, Marko ; Žagar, Marinko ; Ivelja, Tamara

Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni

Izvornik
The 15th International Symposium “Mine Action 2018.” Book of Papers / Adlešič, Đurđa ; Bajić, Milan ; Mateša Mateković, Nataša ; Modrušan, Zdravko ; Njari, Eugen ; Pavković, Nikola ; Trut, Damir - Zagreb : Hrvatski centar za razminiranje - Centar za testiranje, razvoj i obuku (HCR-CTRO), 2018, 25-28

Skup
The 15th International Symposium "Mine Action 2018"

Mjesto i datum
Slano, Hrvatska, 09.04.2018. - 12.04.2018

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Nije recenziran

Ključne riječi
Strong mine presence indicators, DNN, Rule Based System

Sažetak
Suspected Hazardous Area extent is a very important task. It can be done by using advanced computer vision methods and artificial intelligence algorithms on airborne and space imagery in order to extract new information of SHA and to detect indicators of mine presence. We introduce the concept and describe a procedure of strong mine presence indicators detection by using convolutional neural networks and rule-based inference. Also, we propose a recommender system that improves detection quality with interactive relevance feedback. Such a system may also assist in post-processing procedures and classification of indicators after their detection.

Izvorni jezik
Engleski

Znanstvena područja
Geofizika, Geodezija, Informacijske i komunikacijske znanosti

POVEZANOST RADA