Schema for CSV training data labels:

1. id: A unique sequential non-negative integer identifier for the data point.
2. quadkey: A zoom 15 Bing Tile quadkey containing the center of the image.
3. latitude: The latitude of the center of the image in the WGS 84 datum.
4. longitude: The longitude of the center of the image in the WGS 84 datum.
5. num_true: Number of true votes, i.e., votes that this image contains a building.
6. num_false: Number of false votes, i.e., votes that this image does not contain a building.
7. num_unsure: Number of unsure votes, i.e., votes that this image is unclear whether it contains a building.
8. num_error: Number of error votes, i.e., votes that this image has an error or image artifact.

HighResolutionSettlementLayer requires or works with

• Mac OS X or Linux
• and presently only requires Gzip to unpack and ability to access CSV (comma seperated value) files

• CURRENTLY NOT APPLICABLE

• CURRENTLY NOT APPLICABLE

This repo contains a selection from the training, validation, and test data for the [High Resolution Settlement Layer] (https://dataforgood.facebook.com/dfg/tools/high-resolution-population-density-maps) dataset. The current data consists of 9869460 data points with location and rater scores.

The file labels.csv.gz is a Gzip compressed file; to uncompress on unix-like systems run

$ gunzip labels.csv.gz

More information about Gzip, including how to decompress on other systems, can be found at https://www.gnu.org/software/gzip/ .

This will produce a file labels.csv, which is a CSV file with header and the schema provided in the Examples section above or in the README file in more detail.

• Website: https://dataforgood.facebook.com/dfg/tools/high-resolution-population-density-maps
• Facebook page: NA
• Mailing list: NA
• irc: NA

See the CONTRIBUTING file for how to help out.

HighResolutionSettlementLayer is MIT licensed, as found in the LICENSE file.