Тип публикации: доклад, тезисы доклада, статья из сборника материалов конференций
Конференция: Региональные проблемы дистанционного зондирования Земли; Красноярск; Красноярск
Год издания: 2018
Ключевые слова: Deep-learning, Multi-class classification, Polygraphus proximus, Siberian fir, Uav, aerial photography, forest monitoring
Аннотация: Invasion of the Four-eyed fir bark beetle (Polygraphus proximus Blandford) causes catastrophic damage to forests with fir trees in Russia especially in Central Siberia. Based on the latest advances in computer vision and machine learning, we applied a deep convolutional neural network (CNN) to perform a multiclass classification ofПоказать полностьюan UAV-derived imagery aiming at recognition of beetle-induced tree damage categories at the scale of individual fir trees in a mixed forest. The experiment was carried out using RGB images acquired at three research plots on the territory of the state nature reserve "Stolby" (Krasnoyarsk, Russia). The approach is built on the overall architecture of the CNN with six convolutional blocks controlled classification that automatically recognizes 4 categories of fir trees damage with high accuracy (97,53 %).
Номера страниц: 235-241
Издатель: Сибирский федеральный университет