A recent press release states, “Pushpin announced today that it had secured a contract with CoreLogic to automatically extract property attributes from aerial imagery using machine learning. Under the contract, Pushpin leveraged its deep learning platform to extract parcel information including roofing material and number of stories from 3-inch overhead and oblique imagery collected by Nearmap. Currently, CoreLogic relies on analysts to manually assign these attributes to properties. Leveraging machine learning will enable CoreLogic to dramatically reduce costs and increase update frequency for its database of 147 million properties in the United States.”
The release continues, “CoreLogic leverages parcel attributes in estimating property values, which are then used by government assessors and insurance companies. During this project, Pushpin developed training sets and deep learning models for accurately extracting these attributes. Pushpin also developed a plan for operationalizing the algorithms within CoreLogic.”
Mikhail Palatnik, an executive of Product Management at CoreLogic, commented, “Pushpin is one of the only companies applying cutting-edge deep learning techniques to aerial imagery to automatically extract features and identify changes… It has been a pleasure working with Pushpin’s innovative team. Going forward, we expect machine learning to be critical to creating and maintaining accurate property records, which our customers rely on.”
Read more at PRweb.
Image used under license from Shutterstock.com