Very high-resolution urban land cover and land use mapping of U.S. cities

Authors: Gang Chen*, University of North Carolina at Charlotte, Yindan Zhang, University of North Carolina at Charlotte
Topics: Remote Sensing
Keywords: Land cover mapping, very high-resolution, deep learning, urban
Session Type: Virtual Paper
Day: 4/10/2021
Start / End Time: 1:30 PM / 2:45 PM
Room: Virtual 22
Presentation File: No File Uploaded


In this study, we have developed a deep-learning-based framework for mapping urban land cover/use (buildings, roads, parking lots, forest, grassland/shrubland, agriculture, water bodies, barren, and others) at a very high-resolution of 1.0 meter. Applying the framework to the NAIP (National Agriculture Imagery Program) imagery, we have successfully produced land cover and land use maps for 20+ major cities in the United States. Our framework addresses several large-area mapping challenges, such as inconsistent viewing or solar elevation angles, and the lack of ample annotated datasets for model training. We are in the process of expanding the spatial coverage of our product and publishing it via an online platform.

Abstract Information

This abstract is already part of a session. View the session here.

To access contact information login