GID stands for Gaofen Image Dataset, which is a large-scale land cover dataset collected using Gaofen-2 (GF-2) satellite images. It has the advantages of large coverage, wide distribution, and high spatial resolution.

The GID consists of two parts:

1. Large-scale classification set: 150 pixel-level annotated GF-2 images.
2. Fine classification set: 30,000 multi-scale image patches, 10 pixel-level annotated GF-2 images

This dataset collects and re-labels training and validation data for 15 categories based on training and validation images for 5 categories respectively.

Wuhan University released this dataset.