Share Email Print

Proceedings Paper

Optimizing deep learning model selection for angular feature extraction in satellite imagery
Author(s): Poppy G. Immel; Meera A. Desai; Daniela I. Moody
Format Member Price Non-Member Price
PDF $17.00 $21.00

Paper Abstract

Deep learning techniques have been leveraged in numerous applications and across different data modalities over the past few decades, more recently in the domain of remotely sensed imagery. Given the complexity and depth of Convolutional Neural Networks (CNNs) architectures, it is difficult to fully evaluate performance, optimize the hyperparameters, and provide robust solutions to a specific machine learning problem that can be easily extended to similar problems, e.g. via transfer learning. Ursa Space Systems Inc. (Ursa) develops novel machine learning approaches to build custom solutions and extract answers from Synthetic Aperture Radar (SAR) satellite data fused with other remote sensing datasets. One application is identifying the orientation with respect to true north of the inlet pipe, which is one common feature located on the top of a cylindrical oil storage tank. In this paper, we propose a two-phase approach for determining this orientation: first an optimized CNN is used to probabilistically determine a coarse orientation of the inlet pipe, followed by a maximum likelihood voting scheme to automatically extract the location of the angular feature within 7.5° . We present a systematic technique to determine the best deep learning CNN architecture for our specific problem and under user-defined optimization and accuracy constraints, by optimizing model hyperparameters (number of layers, size of the input image, and dataset preprocessing) using a manual and grid search approach. The use of this systematic approach for hyperparameter optimization yields increased accuracy for our angular feature extraction algorithm from 86% to 94% and can be extended to similar applications.

Paper Details

Date Published: 14 May 2019
PDF: 11 pages
Proc. SPIE 10986, Algorithms, Technologies, and Applications for Multispectral and Hyperspectral Imagery XXV, 109861D (14 May 2019); doi: 10.1117/12.2519373
Show Author Affiliations
Poppy G. Immel, Ursa Space Systems Inc. (United States)
Meera A. Desai, Ursa Space Systems Inc. (United States)
Barnard College (United States)
Daniela I. Moody, Ursa Space Systems Inc. (United States)

Published in SPIE Proceedings Vol. 10986:
Algorithms, Technologies, and Applications for Multispectral and Hyperspectral Imagery XXV
Miguel Velez-Reyes; David W. Messinger, Editor(s)

© SPIE. Terms of Use
Back to Top
Sign in to read the full article
Create a free SPIE account to get access to
premium articles and original research
Forgot your username?