Land cover mapping is a semantic segmentation problem: each pixel in an aerial or satellite image must be classified into one of several land cover classes. These classes describe the surface of the earth and are typically broad categories such as “forest” or “field”. High-resolution land cover data (≤1m / pixel) is essential in many sustainability-related applications. Its uses include informing agricultural best management practices, monitoring forest change over time and measuring urban sprawl. However, land cover maps quickly fall out of date and must be updated as construction, erosion, and other processes act on the landscape.
Source:
Large Scale High-Resolution Land Cover Mapping with Multi-Resolution Data
Caleb Robinson, Le Hou, Kolya Malkin, Rachel Soobitsky, Jacob Czawlytko, Bistra Dilkina, Nebojsa Jojic
The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2019
