A deep learning algorithm to detect and classify sun glint from high-resolution aerial imagery over shallow marine environments

2021 ◽  
Vol 181 ◽  
pp. 20-26
Author(s):  
Anna B. Giles ◽  
James Edward Davies ◽  
Keven Ren ◽  
Brendan Kelaher
Keyword(s):  
Deep Learning ◽  
High Resolution ◽  
Learning Algorithm ◽  
Aerial Imagery ◽  
Marine Environments ◽  
Shallow Marine ◽  
Deep Learning Algorithm

scholarly journals Segmentation of seagrass (Enhalus acoroides) using deep learning mask R-CNN algorithm

2021 ◽  
Vol 944 (1) ◽  
pp. 012015
Author(s):  
N A Lestari ◽  
I Jaya ◽  
M Iqbal
Keyword(s):  
Deep Learning ◽  
Industrial Revolution ◽  
Learning Algorithm ◽  
Learning Method ◽  
Shallow Marine ◽  
Enhalus Acoroides ◽  
Deep Learning Algorithm ◽  
Underwater Cameras ◽  
Technological Developments ◽  
Research Procedure

Abstract Seagrass is an Angiosperms that live in shallow marine waters and estuaries. The method commonly used to identify seagrass is Seagrass-Watch which is done by sampling seagrass or by carrying a seagrass identification book. Technological developments in the era of the industrial revolution 4.0 made it possible to identify seagrass automatically. This research aims to apply the deep learning algorithm to detect seagrass recorded by underwater cameras. Enhalus acoroides seagrass species identification was carried out using a deep learning method with the mask region convolutional neural networks (Mask R-CNN) algorithm. The steps in the research procedure include collecting, labeling, training, testing models, and calculating the seagrass area. This study used 6000 epochs and got a measure of value generated by the model of ± 1.2. The Precision value, namely the model’s ability to correctly classify objects, reached 98.19% and the model’s ability to find all positive objects, based on system testing was able to perform recall is 95.04% and the F1 Score value of 96.58%. The results showed that the MASK R-CNN algorithm could detect and segment seagrass Enhalus acoroides.

Deep learning of virtual-based aerial images: increasing the fidelity of serious games for live training

2019 ◽  
Author(s):  
Dean Reed ◽  
Troyle Thomas ◽  
Shane Reynolds ◽  
Jonathan Hurter ◽  
Latika Eifert
Keyword(s):  
Deep Learning ◽  
Serious Games ◽  
Learning Algorithm ◽  
Military Training ◽  
Aerial Imagery ◽  
Aerial Images ◽  
High Fidelity ◽  
Natural Environments ◽  
Game Engine ◽  
Deep Learning Algorithm

The aim of rapidly reconstructing high-fidelity, Synthetic Natural Environments (SNEs) may benefit from a deep learning algorithm: this paper explores how deep learning on virtual, or synthetic, terrain assets of aerial imagery can support the process of quickly and effectively recreating lifelike SNEs for military training, including serious games. Namely, a deep learning algorithm was trained on small hills, or berms, from a SNE, derived from real-world geospatial data. In turn, the deep learning algorithm’s level of classification was tested. Then, assets learned (i.e., classified) from the deep learning were transferred to a game engine for reconstruction. Ultimately, results suggest that deep learning will support automated population of highfidelity SNEs. Additionally, we identify constraints and possible solutions when utilising the commercial game engine of Unity for dynamic terrain generation.

Sign in / Sign up

Export Citation Format

Share Document