Building detection on aerial images using U-NET neural networks

This article presents research results of two convolutional neural networks for building detection on satellite images of Planet database. To analyze the quality of developed algorithms, there was used Sorensen-Dice coefficient of similarity which compares results of algorithms with tagged masks. The masks were generated from json files and sliced on smaller parts together with respective images before the training of algorithms. This approach allows to cope with the problem of segmentation for aerial high-resolution images efficiently and effectively. The problem of building detection on satellite images can be put into practice for urban planning, building control, etc. © 2019 FRUCT.

Авторы
Ivanovsky L.1 , Khryashchev V.1 , Pavlov V.1 , Ostrovskaya A. 2
Редакторы
-
Сборник материалов конференции
Издательство
IEEE Computer Society
Номер выпуска
-
Язык
Английский
Страницы
116-122
Статус
Опубликовано
Подразделение
-
Ссылка
-
Номер
8711930
Том
2019-April
Год
2019
Организации
  • 1 P.G. Demidov Yaroslavl State University, Yaroslavl, Russian Federation
  • 2 People's Friendship University of Russia, RUDN University, Moscow, Russian Federation
Ключевые слова
Antennas; Neural networks; Satellites; Aerial images; Building controls; Building detection; Convolutional neural network; Dice coefficient; High resolution image; Research results; Satellite images; Image segmentation
Дата создания
19.07.2019
Дата изменения
19.07.2019
Постоянная ссылка
https://repository.rudn.ru/ru/records/article/record/38637/