Deep learning was introduced around 2006 as a novel and effective algorithm to train deep neural networks by pre-training one hidden layer at a time using the unsupervised learning procedure for deep belief nets. Deep learning concept became famous in the computer vision community since Deep Convolutional Neural Networks (DCNN), a supervised version of deep learning networks, made a breakthrough by almost halving the error rate of the 2010 Large Scale Visual Recognition Challenge. Deep Convolution Neural Networks has since then rapidly reached out to many industrial applications and other academic areas in recent years as it continues to advance technologies in areas like speech recognition, medical diagnosis, autonomous driving or even the gaming world. Accompanying the rapid development of deep learning techniques is the increased availability of computing equipment (e.g., GPU and hyper computers) and large volume of remote sensing data that has been continuously collected by existing and emerging sensors mounted on various remote sensing platforms. Compared to traditional classifiers (e.g., random forest, support vector machine), DCNN does not need extraction and selection of hand-crafted features. Such advantage, together with its success in the computer vision field has motivated researchers in the remote sensing community to investigate its usefulness for remote sensing image analysis. Therefore, it’s time to take a new look at the existing techniques for processing the remote sensing dataset by taking advantages of recent developments of the algorithm and hardware in deep learning area.
This session aims to provide a platform for remote sensing researchers to share their new research discoveries and thoughts about deep learning applications in remote sensing fields. Example topics include but not limited to:
(1) Deep learning applications for traditional remote sensing tasks such as landcover classifications.
(2) Novel deep learning architectures developed for remote sensing data analysis
(3) Novel framework to trigger the power of deep learning for remote sensing applications by satisfying the demands of massive training dataset by deep learning networks.
(4) Exploration of various deep learning architectures such as Generative Adversarial Networks, Fully Convolutional Networks(FCN), Region-based Convolutional Neural Networks(RCNN), patch-based DCNN for remote sensing data analysis
While this session emphasize remote sensing data, studies using other types of geospatial data such as data collected from social network are also encouraged to join this session.
|Presenter||Tao Liu*, University of Florida, Amr Abd-Elrahman, School of Forest Resources and Conservation, University of Florida, Deep Convolutional Neural Network Training Enrichment using Multi-View Object-based Analysis of Unmanned Aerial Systems Imagery for Wetlands Classification||20||8:00 AM|
|Presenter||Yinan He*, University of North Carolina - Charlotte, GANG CHEN, University of North Carolina at Charlotte, Feng Huang, Yango University; Fuzhou University, Burn severity estimation in a disease affected forest landscape from remotely sensed data: A comparison of empirical, simulation and deep learning models||20||8:20 AM|
|Presenter||Chunxue Xu*, Oregon State University, Bo Zhao, Oregon State University, Deep learning and fake geography: creating satellite datasets with Generative Adversarial Networks||20||8:40 AM|
|Presenter||Fei Hu*, George Mason University, Chaowei Yang, George Mason University, A Spark-based Framework for Big GeoSpatial Raster Data Management and Mining||20||9:00 AM|
|Presenter||Jared Ritchey*, Oregon State University, Bo Zhao, Oregon State University, Jamon Van Den Hoek , Oregon State University, DamageNET: Understanding and Anticipating the Response of a Population to Adverse Events Using Deep Learning||20||9:20 AM|
To access contact information login