CONSTRUCTION OF CLASSIFICATION MODEL OF COASTAL SEDIMENTS IN FUNAFUTI ATOLL OF TUVALU USING DEEP LEARNING
-
- SATO Daisaku
- 摂南大学 理工学部都市環境工学科
Bibliographic Information
- Other Title
-
- フナフチ環礁フォンガファレ島を対象としたディープラーニングによる海岸堆積物分類モデルの構築
- フナフチ カンショウ フォンガファレトウ オ タイショウ ト シタ ディープラーニング ニ ヨル カイガン タイセキブツ ブンルイ モデル ノ コウチク
Search this article
Abstract
<p> Periodic coastal monitoring provides impoartant information on coastal management but it is difficlt to conduct the coastal monitoring in atoll islands that has not enough budget and technical resources. This study constructed the convolution network model which classifies types of coastal sediments by the aerial photographs pictured by drone. In the learning process of model, many set of conditions such as iteration number of learning were executed and compared the learning and test accuracy. The results indicated that increase of middle layer number, iteration number and filter number improves model accuracy. The most accurate model was adapted to the classification of aerial photos. The results showed that the expected result was calculated in sand area. However, the regions of coral gravels had not good result in this study.</p>
Journal
-
- Journal of Japan Society of Civil Engineers, Ser. B2 (Coastal Engineering)
-
Journal of Japan Society of Civil Engineers, Ser. B2 (Coastal Engineering) 76 (2), I_1417-I_1422, 2020
Japan Society of Civil Engineers