Source attribution and licensing
This collection has mixed source terms. license: other and the label
mixed-source-terms describe that fact; they do not create a new blanket license.
Global Streetscapes data and derivatives
The tabular Global Streetscapes source is provided by the NUS Urban Analytics Lab and attributed to Hou et al. (2024), DOI 10.1016/j.isprsjprs.2024.06.023. The upstream repository identifies its license as Creative Commons Attribution-ShareAlike 4.0 International (legal text). Global Streetscapes data and their derivatives in this collection retain that license and source attribution. Changes here include city subsetting, coordinate and capture-sequence joins, explicit scene-label remapping, spatial partitions, exclusions, processed numerical inputs and derived diagnostic labels. The source measurements are attributed to the upstream providers.
The upstream imagery originates from third-party street-view providers; no original street-view photographs are redistributed in this bundle. This bundle does not grant rights to fetch or redistribute photographs under new terms.
IBGE census data and boundaries
IBGE (Instituto Brasileiro de Geografia e Estatística) is the source of the 2022
census-tract income data, census-tract boundaries and FCU boundaries used in the
independent Brazilian analysis. The income table version dated 2026-05-08 retains
2022 as its census reference year. Official download URLs, release information,
file hashes and variable definitions are retained in source_provenance.json
and the archive's IBGE manifests and dictionaries. Joined tables, spatial
aggregations and evaluation results are derived analyses, not official IBGE
findings. Original IBGE notices and applicable source terms remain in force;
this collection does not declare IBGE material to be CC BY-SA or public domain
and does not grant a substitute license on IBGE's behalf.
Category vocabulary, software and documentation
The official Places365 category vocabulary is attributed to the MIT CSAIL Places project; any original notices remain applicable. No Places365 model weights are included.
No new open-source code license or blanket license over the original scripts, reports and documentation is granted by this bundle. Source availability does not itself imply an MIT, Apache, public-domain or other software license. Existing third-party notices and licenses are not displaced. Dependencies remain under their respective licenses.
For source-specific provenance, consult source_provenance.json and
archive_file_descriptions.csv. Keep the attribution and change notices with
redistributed source-derived data.