Training - Tuesday, April 24
Trainings are add-ons and not included in the Symposium Pass registration. Each training session is $25 for USGIF Members and $35 for USGIF Non-Members. Sign up for training during the registration process
Morning Training Sessions
Harvey, Irma, Jose, Maria, and other Trouble Makers—Geolocated Data Use for Hurricane Emergencies – Dun & Bradstreet
This training will examine a use case of providing a combination of proprietary and open source data and geo-coding the data for use by a national agency in the case of national emergencies and disaster relief such as the 2017 hurricanes in Texas and Florida. While this project focused on business data, our client was particularly interested in the “where.” Managing the quality, analysis, manipulation, and integration of that data for both the structured proprietary data and the unstructured open source data required additional work that produced both quality control and analytic building block advantages. A data review produced valuable insights that could be used for future emergency planning, response, and recovery.
Data Science and Machine Learning – IBM
Participants will be led through a series of three labs that build upon one another. The first lab will begin with loading raw data into dashDB for Analytics and interacting with that data from a Jupyter notebook in DSX using python. The second will leverage Spark machine learning on the loaded data to create categorical predictions using pyspark and a supervised learning model and store the results back to the database. The final lab will guide participants in creating a user interface in DSX using RStudio then deploying the completed application into a Shiny application server.
Setting up and Testing NSG and DGIWG Profiles – Open Geospatial Consortium (OGC)
Open Interfaces are crucial to ensure interoperability of GEOINT systems interchanging geospatial information vector data and imager Open interfaces are defined by the Open Geospatial Consortium OGC with open standards such as Web Map Service WMS and Web Feature Service WFS The National System for Geospatial intelligence NSG has taken OGC Standards and has adopted them for the GEOINT community. This tutorial will provide an Introduction about how to setup GEOINT profiles† sing reference implementations and various Mechanism to test these profiles using OGC validation tools Web Validation or locally via a Docker container.
The World in Your Hands in 2 Hours – Thermopylae Sciences + Technology
Google Earth Enterprise (GEE) is a collection of tools that help organizations with imagery and other geospatial data make their information accessible and useful to decision makers and front line users via intuitive visual and fast application. GEE allows users to create private globes from their own geospatial data and to serve these globes and the software from their own data centers, keeping sensitive data behind the firewall with access restricted to authorized users.
Recent Advances in Deep Learning: Fusing Overhead and Ground-Level Views for Remote Sensing – University of Kentucky; UC Berkeley’s International Computer Science Institute
This training is the second in the series, which provides a running update of recent advances in deep-learning methods for automatically extracting geospatial information from ground level and overhead imagery. The focus will be on so-called cross-view methods, which combine these modalities. These methods have been shown to improve accuracy, increase spatial resolution, and reduce the need for manual annotation when compared to methods that only consider a single modality. Applications to scene classification, scene modeling, urban infrastructure assessment, target counting, detection, classification, and segmentation will be highlighted.