Disruptive Technology

Brian Agrow | University of Colorado

Engineers and technologists often develop disruptive technologies before realizing that those charged with developing the regulations and policies for use of the technologies have absolutely no idea what they are dealing with. The result can be technologies run amok with too little oversight, or potentially beneficial technologies that are regulated out of existence.

Attempts at drone policies and regulations have lurched between both extremes. Too often public discourse is dominated by people who are unintentionally or (worse) willfully ignorant. The message is that engineers and technologists must embrace our responsibility to educate policy makers and participate in public discourse. The challenge is to educate future engineers to always be informed on current events, to be capable of articulating technical knowledge to policy makers, and to have the integrity to move industry in the direction of competitive, but responsible, product development.

From Complex Geospatial Data to Answers: Deriving Business Intelligence for the Energy Market from Radar Waveforms

Dr. Daniela Moody | Ursa Space

This talk will demonstrate new approaches employed by Ursa Space Systems Inc. (Ursa) to build custom solutions and extract answers from Synthetic Aperture Radar (SAR) data. These alternative data sets enable novel analyses and interpretation of geospatial multi-modal data. For example, Ursa leverages radar based satellite imagery to measure thousands of crude oil storage tanks every week. With this data, Ursa is able to compare, correlate, and analyze trends, patterns and anomalous changes over time in the frame of the energy and financial markets. This presentation will demonstrate examples of these analyses and talk about how Ursa’s users find qualitative and quantitative value in derived geospatial business intelligence data.

Applications of Hyperspectral Imagery in Precision Agriculture

Dr. Alex Melnitchouck | BASF

This presentation will cover existing and potential applications of hyperspectral imagery in the agricultural industry, as well as the tools for data management and analytics. Using practical examples, we will discuss existing and future sources of hyperspectral data, show advantages of hyperspectral technology, and outline difficulties that data users can encounter while applying hyperspectral imagery in Precision Ag.

SAR Analytics & Data Fusion Driving Insights into Illegal Logging in the Amazon

Rebecca Lasica and Alberto Meroni | Harris

Using SAR data and analytics to answer questions was once a complex and labor-intensive process. Visual workflow development and automated SAR analytics have removed the barriers of complexity. This presentation will introduce SAR analytics innovations in the context of multi-temporal change detection and data fusion related to illegal logging in the Amazon – complete with a live demo!

Bringing HSI to the Mainstream

Richard Zacaroli | Corning

Until recently, hyperspectral imaging systems were quite large and prohibitively expensive for practical application to precision agriculture and other commercial uses. They also produced a tremendous amount of data that was difficult, if not impossible to manage for large scale collects using conventional PC devices. To date, over sixty vegetation indices have been developed and proven for farm and vineyard management. The introduction of new crop and application-specific indices is accelerating. The ability to utilize the wide range of available indices has grown beyond conventional multispectral imagers. In addition, a new class of indices will use the entire spectra, and not just a few selected spectral bands.

A new generation of hyperspectral imaging sensors and turnkey airborne systems is now available that are smaller and lighter for integration with small UAS, and cost little more than higher-end six and eight-band multispectral imaging systems. Depending upon each use case, the user may select to collect the entire hyperspectral data cube, or only the spectral bands needed to produce the desired indices and data products using ENVI® software and web-based image processing services.

Collecting Spectra for Image Analysis

Gary Fager | Malvern Panalytical

Remote sensing of the Earth is a process that involves many trade-offs and challenging obstacles when collecting data for analytical processes. One technique that can assist the analyst or the calibration specialist is to use instrumentation deployed to the field to gather “pristine” spectral data without the confounding effects of the atmosphere, etc. or to perform vicarious calibration of the air or space borne remote sensing instrument. This talk will discuss the ways that a field spectrometer can assist with the process on both the front and the back end.

Use Multispectral Imaging to Boost Crop Health: From Data Collection to Analysis

Dr. Manal Elarab | MicaSense

Drone-based multispectral imaging is quickly becoming a fundamental instrument in agriculture research centers and major agriculture operations. Growers are incorporating imagery into their decision support systems, and researchers use it as a source of data in various agriculture research fields (viticulture, field crops, phenotyping, wetland ecology etc.). However, the key to its ongoing success lies in understanding the different needs of growers and researchers, and determining how to provide solutions that each group can use effectively. This presentation explores industry application examples and discusses what must be created in the future to support growers and researchers as they work to advance precision agriculture.

Seeing Through the Haze: Removing Effects of the Atmosphere to Improve Geospatial Data Products

Marsha Fox | Spectral Sciences

Atmospheric effects, including transmission losses, aerosol scatter, and ground scatter, may adversely affect the radiometric quality of imagery taken from any altitude - from unmanned aerial vehicles to satellites. Atmospheric correction uses in-scene and model information from a sensor to remove these effects. The accessibility of massive quantities of earth imagery, like near-daily SmallSat imagery of many locations, and widely available of UAV systems for periodic local measurements, opens up new opportunities for analytics that can benefit from atmospheric correction. Examples include geospatial data products like surface material classification and mapping, change detection, and data mining algorithm training data and results. This presentation will discuss the value of correction to current and future geospatial applications and review techniques for various platforms and uses.

Demystifying Geo.AI

Omar Maher | Esri

Machine Learning (ML), a branch of Artificial Intelligence (AI), is a disruptive force that is changing how today’s industries are gaining new insight from their data. Within the geospatial industry, this insight is quickly moving from “information about information” to “answers from data”. We will explore how Esri leverages ArcGIS to support cross-platform ML that delivers definitive geospatial data – unlocking the potential of every pixel.

Cross-domain AI Models for Defense and Commercial Enterprises

Carrie Hernandez | Slingshot

Slingshot Aerospace is an advanced analytics company enabling enterprise decision makers to execute mission tasks more effectively by transforming vast amounts of complex, often unstructured remote sensing data into actionable insights. Slingshot’s AI-fueled, cloud-native platform is purpose-built to rapidly assimilate, contextualize, and analyze multimodal optical, SAR, and RF data at scale while remaining agnostic to source and inclusive of alternative and commercially available data. In tandem, the company is integrating their pruned machine-learned models onto select AI-capable chips for embedded edge compute onboard drones and satellites. This presentation will highlight Slingshot’s ability to provide greater context and depth around situational awareness with its automated object detection, monitoring, and tracking capabilities designed specifically for defense, disaster response, and insurance customers.

Scaling Deep Learning for Multispectral Imagery

Dr. Hanlin Tang | Intel

Intel’s AI Products Group has been at the forefront of developing novel algorithms and deploying deep learning at scale. Remote sensing data has unique characteristics, such as hyperspectral bands, yet existing deep learning algorithms were designed for academic RGB datasets. In this talk, we propose new deep learning architectures that use attention mechanisms to achieve state-of-the-art results on building detection. We demonstrate that our model is significantly more robust to individual band noise because it dynamically allocates attention away from the affected bands. We end by sharing how to scale deep learning algorithms to real-world deployments for enterprise and government agencies.

Synthetic Labeled Data for Application of Deep Learning to Remote Sensing Imagery

David Gorodetsky | Harris

Convolutional Neural Nets (ConvNets) have become the de facto standard solution for computer vision problems. In particular, models incorporating Deep Learning are routinely the top performers – often by sizable margins – for tasks in object identification and localization, scene understanding, image segmentation, and video action recognition. While these methods were initially developed using natural images from social media sources, they have also been applied with great success to data from airborne and satellite imagery. Nevertheless, adoption of these techniques by the remote sensing community has been hampered by the requirement for extremely large collections of labeled training data, which are rarely available and prohibitively expensive to acquire.

In an effort to overcome this labeled data burden, Harris has been exploring the use of synthetic data from physics-based models and simulations, which can produce radiometrically accurate labeled data in virtually unlimited quantities. When combined with advanced techniques such as Generative Adversarial Networks (GANs) and Domain Adaptation, it may be possible to enable model training and transfer learning with highly limited labels from the real source data.

Multi-temporal and Multi-spectral Data Fusion Applications for Managed Agro-ecosystem Dynamics

Dr. Narcisa Pricope | UNC Wilmington

Managed agro-ecosystems worldwide (especially those located in drylands) are facing increased pressures from shifting climatic and socio-economic patterns and conditions. Earth observation (EO) data collected at different spatio-temporal scales in concert with socio-economic household surveys can help shed important light into people’s coping mechanism and potential adaptation strategies.

This presentation will focus on multi-institutional and interdisciplinary work that leverages extensive EO data, unmanned aerial vehicle photogrammetry and biophysical and socio-economic survey data across three countries in the world’s largest transfrontier conservation area in southern Africa. The presentation will show that multi-temporal, multi-spectral and multi-scalar EO data are key contributors to disentangling complex relationships in managed agro-ecosystems.

Open SAR Imagery and Security Awareness

Paolo Pasquali | sarmap

The recent broad availability of open satellite SAR data, like those from the ESA-EU Sentinel-1 mission, is not only beneficial for civil land applications. Despite their medium (~20m) resolution, the regular and continuous acquisition plan and the high temporal resolution make these data interesting also for security applications, where very high spatial resolution requirements are normally expected.

This presentation will show a number of real examples demonstrating the type of information that can be derived from open SAR data time series analysis to increase awareness on different security topics, including oil smuggling, construction of airstrips, evolution of battles and jamming activities.

Mining Insights and Analytics from Daily Imagery

Annie Neligh | Planet

Most organizations that leverage satellite imagery are drowning in pixels, overwhelmed with the amount of data to sift through and the manual extraction of relevant insights. Often, the imagery is outdated by the time it’s received or provides an incomplete dataset, hindering timely, data-driven decisions. Multi-temporal analysis and predictive analytics are changing the game. Planet will deep dive into their latest analytic offering of building and road detection to show how applying machine learning to our deep stack of imagery enables timely insights over broad areas of the globe.

FEMA’s Use of Geospatial Data and Analytics for Emergency Response

Christopher Vaughan | FEMA

In 2017, the United States experienced some of the most significant disasters in its history in terms of their social and economic impact. FEMA is the key federal agency tasked with addressing disasters. In recent years they have made it a strategic priority to develop GIS data and cutting-edge technologies that now enable them to more effectively support communities before, during, and after disasters. In this presentation, you will see demonstrations of the tools now available from FEMA and learn how they are being used around the nation.

Changing Situational Decision Making Using Multi-source Aerial Imagery and Analytics for Catastrophic Response and Insurance Claims

Jesse Kallman | Airbus Aerial

Airbus Aerial offers customers the ability to combine insights derived from Airbus’ massive, global archive of premium satellite imagery with the option to task satellites and operate manned and unmanned aerial missions for even more detailed information about their assets, customers, etc. Understanding where significant change is happening across thousands of miles scattered across an entire city requires an approach that optimizes the cost and time to deliver insights through best in class analytics.

From automatically triaging total loss structure damage using EO satellites, to fine damage classification using drones, to tracking flood levels with radar satellites, comprehensive damage assessment is being accomplished by fusing these data types together. Airbus Aerial recognizes an opportunity to apply conventional remote sensing and machine learning analytics to multi-source aerial imagery for making critical decisions.

Earth Science 2.0

Dr. Sheldon Drobot | Harris

In today’s interconnected world, satellites provide critical information affecting national well-being. Over the last few years, the nature of space-based observation has begun to shift along two important pathways. First, rapid advances in manufacturing technology are opening new prospects for low-cost constellations to improve temporal monitoring. Second, further developments in remote sensing technologies are creating the possibility for hyperspectral information.

Hyperspectral remote sensing collects a complete spectrum of data for each pixel, which in turn opens new avenues for geophysical interpretation of narrowband characteristics. Most features on Earth have characteristic responses along the electromagnetic spectrum and by parsing the spectrum into discrete blocks, we can precisely identify information in ways that were not possible with multispectral data. With this information, we can better monitor, understand, and predict the evolution of agriculture, forestry, marine resources, emergency management, and other fields. Combined, this will lead to a “Earth Science 2.0” revolution in observing capacity. This presentation will discuss how hyperspectral constellations can provide cutting edge information for nations and businesses. It will focus on next generation technologies, concepts, and plans for acquiring hyperspectral data from space, and what you need to do to be prepared to take advantage of this coming revolution.

Delivering Effective GEOINT Solutions to Disaster Relief

Jessica Hulsey | BAE

With the frequency and severity of natural disasters on the rise, it’s more important than ever to put GEOINT into the hands of response planners and operators. BAE Systems Geospatial eXploitation Products™ (GXP®) and Harris are bringing our geospatial and analytic expertise to develop use-case driven prototypes designed to provide value for even the most novice user. See how GXP Online and ENVI spectral analytics come together to provide robust data processing and visualization capabilities to aid disaster response.

Taming the Information Rivers Before Swimming in the Data Lake

Dr. Stanley Grossman | George Mason University

Current intellectual concentration is on the act of analytics ignoring the source of the analytical data. This briefing is a discussion of the importance of foundational data and observations, the value of what is being missed, and how these data might be generated.

Situational Awareness Tools (SAT): Geospatial Models Supporting Marines

Dr. Douglas Way | Radiant Solutions

The Marines are often challenged to deploy in global environments that have diverse geography, culture, social, political, economic, environmental, and governance conditions. While databases constantly evolve, geospatial modeling assists in providing agile intelligence. Situational Awareness Tools (SATs) were developed to provide subnational insight such as market access (remoteness), economic fabric, societal support infrastructure, travel time, helicopter landing zones (HLZs), drop zones, aerial concealment, ambush opportunities and vulnerabilities, remote havens, vehicle mobility, sources of construction materials, health care access and exposure to natural hazards. These models at a 30 or 10-meter resolution support rapid response providing analysts with first order products that serve as a foundation for further analyses.

Tactical GEOINT: New Opportunities for Mission Success via Time-Dominant Exploitation

Stuart Blundell | BlackSky

Advances in analytical methods and increasing diversity in commercial GEOINT sources are enabling the tactical use of GEOINT for information superiority and decision advantage. The real-time demands of Tactical GEOINT entail heavy reliance on Time-Dominant Exploitation (TDE) through automation, including the means for parsing and correlating near real time multi-source detections to create longitudinal statistical models. The deluge of commercial and open-source data available for TDE are forcing the adoption of new paradigms in machine-based analysis and interpretation. These changes are having an especially acute impact on the creation of patterns-of-life and normalcy models. In TDE, the objective is to orchestrate the source collection and data science such that real-time analysis and model-based time-dependent predictions unite to enhance awareness. BlackSky’s asynchronous geospatial services architecture redefines the traditional TCPED (Tasking, Collection Processing, Exploitation, and Dissemination) approach to geospatial exploitation and establishes a new paradigm for Tactical GEOINT.

Spectral Imaging for Cultural Heritage - Redaction and Extraction

Dr. Fenella France | Library of Congress

The expanded utilization of spectral imaging for non-invasive assessment of cultural heritage has led to applications that enable non-invasive chemical identification of historic materials, the recovery of obscured and degraded non-visible information, and the capacity to monitor non-visible changes in fragile materials to optimize conditions for storage and display. These are important considerations for fragile historic materials and allow new discoveries to be made; damaged lost Aztec writings, original text changes by founding fathers, how object and documents were created, provenance of materials, and identification of pigments and inks to reveal different authors and creators.

Spectral analysis advances the capability for identification of colorants, inks, and substrates through their specific spectral response, aiding provenance of historic artifacts. Monitoring deterioration or changes due to exhibit and other environmental conditions is critical for the longevity of cultural heritage objects and spectral curve analysis is a critical feature of this preservation application. Image processing from at-risk heritage materials range include pre-Columbian vessels, the original rough draft of the Declaration of Independence and 21st century political cartoons, allowing the extraction of fresh information about these materials, and world cultures, without damaging our history through other invasive methods of analysis.

Remote Sensing and Data Analytics Opportunities from the International Space Station

Dr. Etop Esen | CASIS

The location of the International Space Station (ISS) in low Earth orbit (LEO) offers a unique vantage point for Earth and Space observation. Recent and planned installations of external platforms on ISS provide new opportunities for simultaneous hosting of multiple payloads for new sensor development, testing, and data handling. Additionally, the ISS has capabilities to deploy remote sensing free-flyers into LEO. This presentation will cover the variety of existing and new, customizable configurations for sensor mounting, data acquisition and handling, and how the ISS can facilitate development of new data analytics and products for targeted industries and applications such as energy, marine and agriculture.

Geospatial Analytics at Xcel Energy

Gray Byers | Xcel Energy

Xcel Energy is an industry leader in the utilization of drone technology for utility asset inspection and is currently developing a robust analytics strategy for large asset data sets, utilizing artificial intelligence and other advanced analytics to facilitate workflow. This presentation will focus on the current state of Xcel Energy’s drone program as well as its future vision for the program and analytics strategy for remote sensing data collection within operations.

Remote Sensing and IoT: Changing the Way Businesses Compete

Eric Dixon | Harris

The operations of large retailers and property owners are affected by weather such as patron safety, snow removal, power washing, and parking lot sweeping. Retailers and property owners often become anxious when a weather event occurs near their buildings. Real-time weather information from each location can reduce this anxiety and be extremely valuable especially during extreme weather. Some of the benefits of real-time weather information include:

• Awareness of quick changing weather conditions that are not forecasted (wet or snow-covered parking lots).
• A resource for management to optimize snow removal, manage operations, power washing, parking lot sweeping, etc.
• Reduce parking lot snow removal fraud
• Ability to understand current progressing weather which may affect store operations

Blockchain is the Answer, But What was the Question?

Ivo Banovsky | CloudEO

Blockchain solves two critical issues facing the earth observation industry. First, improving open geospatial data veracity since it is difficult at best to determine the origin of an open geospatial dataset and which alterations, if any, have been performed. The second is to lower the entry and access barriers to the earth observation industry for the broader public?

The CLEO Blockchain Network (CBN) Foundation utilizes blockchain technology for smart contracts and to establish chains of custody for government, commercial, and open geospatial data. This presentation provides an overview of the core qualities of blockchain in relation to geospatial data, and how the CBN Foundation platform can be utilized to perform data analysis techniques on the geospatial datasets it receives, encode them to the blockchain to provide an extra layer of protection against manipulation, and establish a chain of custody.

OneAtlas: From Imagery and Data to Knowledge and Decision-Making

Jessi Dick | Airbus Defence and Space

Designed with you and for you, OneAtlas is a unique collaborative environment enabling you to easily access constantly updated premium imagery, perform large-scale image processing, extract industry-specific insights, and benefit from Airbus assets to develop your solutions. Learn more about the soon-to-be launched data and analytics services provided by Airbus and receive exclusive insight on the future plans for this new platform.

DigitalGlobe’s GBDX Platform with Harris Analytics

Adam O’Connor | DigitalGlobe

Harris and DigitalGlobe have partnered to deploy Harris’ cutting edge geospatial analytics with DigitalGlobe’s satellite image library on the GBDX cloud computing platform to extract spatial information. This unique combination of a massive 100 petabyte satellite image library with Harris analytics allows users to solve location intelligence problems at massive scale. This presentation will provide an overview of how Harris analytics including ENVI advanced image processing and MEGA machine learning algorithms can be used on the GBDX platform in order to derive actionable insights that improve business processes and enhance mission success.

Harris Vision for Building and Deploying GEOINT Solutions

Dan Conser | Harris

To provide our clients actionable information when and where it is needed, Harris is evolving existing legacy capabilities to modern cloud platform solutions. By fostering a culture of client-centric solution development, Harris will solve our client’s most important and challenging problems while incorporating new subscription and consumption business models that better align with our client’s revenue models.