East View Geospatial (EVG) continues to enhance the accuracy of automated feature identification using its newly-developed training data sets in supervised machine learning applications. The early results pertain to automated recognition of building structures in an ongoing pilot project in Papua New Guinea (PNG).
“Our goal is to create a state-of-the-art process that produces the highest quality training data available for the users and developers of supervised machine learning technology,” says Rod Buhrsmith, EVG business development. “In just a few months, we have made significant progress and expect to push the accuracy even higher.”
EVG initiated the PNG pilot early this year with multiple organizations actively engaged in enhancing and calibrating big data algorithms. Most supervised machine learning applications in geospatial currently entail either automated extraction of information from massive data sets or management and tracking of high-velocity, high-volume data streaming from numerous remote sensing platforms.
In its early phases, the pilot has concentrated on developing foundational training data sets tailored to identification of transportation, hydrology, infrastructure and land use/land cover in half-meter satellite imagery. EVG and its partners are now expanding their research to include imagery and streaming video from other sources, including aircraft, unmanned aerial systems (UAS) and terrestrial sensors.
“Our role in this space is to become the authoritative source of training data, either custom made or archived in a commercial library,” says Mark Knapp, EVG product development. “We have the data sourcing expertise and MilSpec production capabilities to accommodate any geospatial application.”
In the next phases of the PNG pilot, EVG will create custom training data sets for specific supervised machine learning applications and geographic areas in parallel with development of off-the-shelf data for on-demand purchase. This geospatial data will be drawn from the EVG archive, which includes imagery, maps and charts, as well as topographic, 3D, DEM, GIS and vector data sets. Much of the data is locally sourced in parts of the world not accessed by any other commercial cartographic organization. Areas include Russia, China, Iran, Former Soviet Union, Middle East, East Asia and North Africa.
About East View
East View was founded in 1989 and is headquartered in Minneapolis. East View is comprised of East View Information Services, East View Geospatial and East View Map Link. East View maintains thousands of supplier/publisher relationships throughout the world for maps and geospatial data and Russian, Arabic and Chinese-produced social and hard science content. East View manages a data center, library and warehouse in Minneapolis where it hosts and stores dozens of foreign language databases, hundreds of thousands of maps and atlases and millions of geospatial, Russian, Chinese and Arabic metadata records.