MUSA GIS Day 2017: Computer Vision and Remote Sensing in Spatial Analysis

On November 15, Penn IUR and the Masters of Urban Spatial Analytics (MUSA) program celebrated GIS Day, a national day addressing the importance of geospatial analysis. The event brought together academics and practitioners whose presentations showcased innovations of Geographic Information Systems (GIS) in research and real-world applications. This year’s annual GIS Day celebration focused on the cutting edge “deep learning” technology that is revolutionizing the field of remote sensing. Deep learning technology can take a picture from space and count the number of cars parked outside a stadium during Sunday’s big game, or enable a drone to see an obstacle and avoid it in real time. Additionally, a prevailing theme discussed throughout the event was the potential adverse impacts that technology can have on cities.

Penn IUR Faculty Fellow John Landis, MUSA Faculty Advisor and Crossways Professor of City and Regional Planning, School of Design, opened the event by discussing the impact technology has had on cities, both in terms of advancements and potential adverse effects. Though the significance of artificial intellegence (AI) and technology is clear, its future impacts are not; if implemented responsibly, such technologies could change the way urban design and management technology look and function.

Chris Holmes, SVP, Product Architecture at Planet, discussed major trends prevalent in the current explosion of information: consumer electronics; the ability to launch satellites inexpensively into space; the success of drones as photographic platforms adapted for businesses and individuals; and the fruition of the Cloud and its ability to make data more accessible. Holmes discussed the seemingly limitless possibilities that come with the ability to identify and utilize recorded images. Still, many questions remain:  How do we get to a stage where all the information is accessible to everyone? How do we make these massive data sets of information sized and digestible for an everyday consumer?

Abhishek Gaur, Deep Learning Engineer, Neurala, Inc, emphasized the importance how we implement deep learning techniques. Deep learning has been around for a long time but with improved hardware and the emergence of greater data availability, it now offers new solutions for present urban challenges. The most pressing problem with extended information technology is data privacy, according to Gaur, and many people remain uncomfortable disclosing private information on external sources such as the Cloud.

Rob Emanuele, Vice President of Research and Technical Lead on the GeoTrellis Team at Azavea, discussed how deep learning techniques are applied to geospatial imagery. GeoTrellis, an open source, geographic data processing library designed to work with large geospatial raster data sets in the Scala language, allows more effective analysis and use of satellite imagery. This program allows people to analyze the mass of widely available data in a more streamlined and user-friendly way.

Following the presentations, Penn IUR Emerging Scholar Ken Steif, MUSA Program Director and Lecturer, City and Regional Planning, School of Design, moderated a conversation about the relationship between technology, the public and policymakers. The discussion touched on the challenge of technology development outpacing government regulation and the importance of the government addressing privacy concerns. Participants also discussed to how to best utilize the large amounts of data sourced through deep learning technologies. To learn more, watch the full video online. 

Watch the Video: Penn GIS Day, Part 1

Watch the Video: Penn GIS Day, Part 2