DEVELOPMENT OF A SEDIMENT PARTICLE SIZE PREDICTION METHOD FOR TIDAL FLAT BASED ON IMAGE ANALYSIS USING DEEP LEARNING

DOI Web Site 1 Citations 4 References

Bibliographic Information

Other Title
  • 機械学習を用いた画像解析による干潟の粒度推定手法の開発

Description

<p> Our study aimed to develop a simple particle size estimation method for tidal flats and to verify its applicability for sediment classification of tidal flats. We conducted 1) the survey of the particle size distribution of tidal flats and photographing of the surface layer at 51 points located at the mouth of the Saba River in Yamaguchi Prefecture and 2) the estimation of the particle size by image analysis using machine learning from the surface photographs. The results showed that the tidal flat surfaces with mud content of less than 20% and those with mud content of 20% or greater could be discriminated from the images with an accuracy of approximately 73%. Next, tidal flat images were classified into four categories of mud contents using machine learning; less than 15%, 15% to 25%, 25% to 35%, and 35% or more. The percentages of correct images in each category were 72%, 48%, 45%, and 75%, respectively. These results suggest that the machine learning image analysis used in this study can improve the accuracy of the discrimination when the partition of the mud content is set to be coarser.</p>

Journal

Citations (1)*help

See more

References(4)*help

See more

Keywords

Details 詳細情報について

Report a problem

Back to top