AWS re:Invent 2018 Geospatial Talks
Talks, sessions and workshops that may be of interest to those working with geospatial data. PRs accepted!
There are unique challenges to building highly accurate models that detect small objects in aerial and overhead imagery. In this chalk talk, we dive deep into using convolutional neural networks (CNNs) with Amazon SageMaker in order to build and train aerial object detection models. We build advanced models using AWS public datasets, such as SpaceNet and LandSat, as we work with DigitalGlobe's GBDX Notebooks.
Across the commercial and public sectors, companies are working with large geospatial datasets. In this session, we review how to use various services, including Amazon S3, Amazon Glacier, Amazon Athena, AWS Batch, and AWS Step Functions to store, process, and get insights into your large geospatial datasets.
In a short space of time, fast.ai has become a popular Deep Learning library, driven by the success of the fast.ai online Massive Open Online Course (MOOC). It has allowed SW developers to achieve, in the span of a few weeks, state-of-the-art results in domains such as Computer Vision (CV), Natural Language Processing (NLP), and structured data machine learning. In this chalk talk, we go into the details of building, training, and deploying fast.ai-based models using Amazon SageMaker.
AWS hosts a variety of public data sets that anyone can access for free. Previously, large data sets such as satellite imagery or genomic data have required hours or days to locate, download, customize, and analyze. When data is made publicly available on AWS, anyone can analyze any volume of data without downloading or storing it themselves. In this session, the AWS Open Data Team shares tips and tricks, patterns and anti-patterns, and tools to help you effectively stage your data for analysis in the cloud.
WPS315 - National Geospatial-Intelligence Agency: Changing the Way the Intelligence Community Moves Data
In this session, we feature the U.S. National Geospatial-Intelligence Agency (NGA), a key stakeholder and sponsor for the new AWS Secret Region, which supports workloads up to the Secret U.S. security classification level and is readily available to the U.S. Intelligence Community (IC). NGA uses AWS Snowball Edge to support War Fighter, utilizing imagery from NGA’s Open Data Store and implementing geospatial applications on the edge. AWS Snowball Edge allows NGA to directly support its mission, providing products and services to decision makers, warfighters, and first responders when they need it most. Enabling the edge changes NGA’s ability to share critical resources, data to facilitate user access meets NGA’s mission needs, and support the IC and Department of Defense as a whole.
In this chalk talk, we discuss how to design a data lake, and how to permission different groups and applications to access and analyze datasets. Learn from subject-matter experts about a variety of AWS technologies for populating your data lake, monitoring new ingestion, and processing data for meaningful analysis. We examine considerations for structured data, such as relevant database engines with geospatial support, as well as considerations for unstructured data in the form of object storage. In addition, we address how to protect and secure data based on an organization’s needs.
Jupiter is a cloud-native company that delivers hyperlocal environmental information in a changing climate, primarily using AWS Batch. Through AWS Batch’s capability to execute thousands of scientific modeling jobs while managing scale and cost, Jupiter scientists can focus on data analysis and developing sophisticated machine learning (ML)-based applications to support private sector and local municipality customers; AWS Batch takes care of the rest. In this chalk talk, we demonstrate how AWS Batch, through managing resource provisioning and scheduling, enables flexibility across changing requirements to allow various modeling applications to run quickly and at scale.
What’s the shakeup in Silicon Valley? Join us as we investigate global subduction zones, highlighting and plotting areas with the deepest earthquakes. Using Amazon Athena, Amazon Redshift, and Matillion ETL for Amazon Redshift, we prepare a semistructured geospatial dataset from the International Federation of Digital Seismograph Networks for visualization. Learn how to build a best-practice architecture using Athena to read and flatten Amazon S3 data, Matillion ETL to perform the more complex data enrichment, and Amazon Redshift for aggregation, before handing off the data to Amazon QuickSight for visualization. This presentation is brought to you by AWS partner, Matillion Limited.
Join us for a deep dive on building a searchable image library using Amazon Rekognition. We walk though creating a search index for objects and scenes so you can quickly retrieve images using labels created from automatic metadata extraction. Also learn how to use AWS Lambda to automatically maintain your image library.
Get hands on with serverless data processing at scale. In this session, we use Landsat 8 satellite imagery to calculate a Normalized Difference Vegetation Index (NDVI) across multiple points of interest in the world using the GeoTIFF data across multiple spectral bands.
As a leader in agriculture technologies and services, Bayer is using technologies such as unmanned aerial vehicles (UAV), satellite imagery, and sensor data from multiple sources to generate real time insights. Over 300 data sources are ingested into their open source HPC geospatial platform to generate on average 100M API calls per day. The platform is used to provide real-time visualization and computational analysis to Bayer’s internal research community, partners, and is licensed to third-party applications to provide insights relevant to high-yield production of crops. In this session, Mendez-Costabel discusses how Bayer transitioned from on-premises packaged software architecture to open-source software and cloud services from AWS to build a modern, scalable, high-performance, open-source app on AWS. Learn about the open-source application architecture and AWS services used. Learn how the computing environment has changed the way that Bayer is performing R&D projects, and how the move to a modern architecture has enabled Bayer’s customers to gain insights that are transforming their businesses.
Learn how to use Amazon SageMaker and other AI/ML services on AWS to build predictive data models for life sciences. In this workshop, you’ll learn how to incorporate diverse data types ranging from sensor data to imagery, into machine learning models. At the end of this workshop, you’ll be able to deploy these models to monitor real-world scenarios in life sciences.
In this session, learn how an AWS HPC customer in the aerospace engineering segment migrated key parts of their computer-aided engineering (CAE) simulation and visualization applications to AWS to improve infrastructure redundancy for a robust process and user experience, achieving a system speed-up of 18X.
Streaming data ingestion and near real-time analysis gives you immediate insights into your data. By using AWS Lambda with Amazon Kinesis, you can obtain these insights without the need to manage servers. But are you doing this in the most optimal way? In this interactive session, we review the best practices for using Lambda with Kinesis, and how to avoid common pitfalls.
AWS offers a wide selection of compute platforms. In this session, we highlight key platform features of different Amazon EC2 instance families, and provide a framework in which to choose the best compute resource (including Amazon EC2 Instance, AWS Fargate Container, and AWS Lambda function) for your workloads based on metrics and workload profiles. We also share best practices and performance tips for getting the most out of your Amazon EC2 instances to help you reduce unnecessary spending and improve application performance.
Amazon Kinesis Video Streams makes it easy to capture live video, play it back, and store it for real-time and batch-oriented ML-driven analytics. In this session, we first dive deep on the top five best practices for getting started and scaling with Amazon Kinesis Video Streams. Next, we demonstrate a streaming video from a standard USB camera connected to a laptop, and we perform a live playback on a standard browser within minutes. We also have on stage members of Amazon Go, who are building the next generation of physical retail store experiences powered by their "just walk out" technology. They walk through the technical details of their integration with Kinesis Video Streams and highlight their successes and difficulties along the way.