rpradana36/campus-svi-availability

Crowdsourced vs. proprietary street view coverage around Indonesian university campuses — Database

Mohammad Raditia Pradanaᵃ'ᵇ'*, Jarot Mulyo Semediᵃ'ᵇ

ᵃ SPARC (Spatial Modeling & Analysis Research Cluster), Universitas Indonesia, Depok, 16424, West Java, Indonesia ᵇ Department of Geography, Faculty of Mathematics and Natural Sciences, Universitas Indonesia, Depok, 16424, West Java, Indonesia

Grid-cell-level and point-level street view imagery (SVI) coverage data for 40 Indonesian university campuses, built from Google Street View and Mapillary metadata. The dataset supports research on the spatial, temporal, and platform-driven inequality of street-level imagery availability across campus environments — coverage, positional agreement between platforms, temporal depth, contributor concentration, and spatial clustering (Moran's I).

For each campus, raw per-image metadata is aggregated onto a uniform 20 m grid, giving a directly comparable "openness" surface across institutions of very different size, shape, and location.

Campuses

40 campuses across Java, Sumatra, Kalimantan, Sulawesi and Bali, defined in reference/campus_registry.csv and boundary files under boundaries/.

Slug (campus_id)	Display name	Full name	City	Province	Island
ipb	IPB	IPB University (Dramaga)	Bogor	West Java	Java
itb_ganesha	ITB Ganesha	Institut Teknologi Bandung — Ganesha	Bandung	West Java	Java
itb_jatinangor	ITB Jatinangor	Institut Teknologi Bandung — Jatinangor	Sumedang	West Java	Java
itera	ITERA	Institut Teknologi Sumatera	South Lampung	Lampung	Sumatra
its	ITS	Institut Teknologi Sepuluh Nopember	Surabaya	East Java	Java
telkom_uni	Telkom University	Telkom University	Bandung Regency	West Java	Java
ugm	UGM	Universitas Gadjah Mada	Sleman	DI Yogyakarta	Java
ui_main	UI	Universitas Indonesia — Depok	Depok	West Java	Java
uii_kaliurang	UII	Universitas Islam Indonesia — Kaliurang	Sleman	DI Yogyakarta	Java
um	UM	Universitas Negeri Malang	Malang	East Java	Java
umy	UMY	Universitas Muhammadiyah Yogyakarta	Bantul	DI Yogyakarta	Java
unair_b	UNAIR Campus B	Universitas Airlangga — Campus B	Surabaya	East Java	Java
unair_c	UNAIR Campus C	Universitas Airlangga — Campus C	Surabaya	East Java	Java
unand	UNAND	Universitas Andalas	Padang	West Sumatra	Sumatra
unbraw	UB	Universitas Brawijaya	Malang	East Java	Java
undip	UNDIP	Universitas Diponegoro — Tembalang	Semarang	Central Java	Java
unej	UNEJ	Universitas Jember	Jember	East Java	Java
unesa_1	UNESA Campus 1	Universitas Negeri Surabaya — Campus 1	Surabaya	East Java	Java
unesa_2	UNESA Campus 2	Universitas Negeri Surabaya — Campus 2	Surabaya	East Java	Java
unhas	UNHAS	Universitas Hasanuddin	Makassar	South Sulawesi	Sulawesi
unila	UNILA	Universitas Lampung	Bandar Lampung	Lampung	Sumatra
unimed	UNIMED	Universitas Negeri Medan	Deli Serdang	North Sumatra	Sumatra
unj	UNJ	Universitas Negeri Jakarta	East Jakarta	DKI Jakarta	Java
unja	UNJA	Universitas Jambi	Muaro Jambi	Jambi	Sumatra
unlam	ULM	Universitas Lambung Mangkurat	Banjarbaru	South Kalimantan	Kalimantan
unmul	UNMUL	Universitas Mulawarman	Samarinda	East Kalimantan	Kalimantan
unnes	UNNES	Universitas Negeri Semarang	Semarang	Central Java	Java
unp	UNP	Universitas Negeri Padang	Padang	West Sumatra	Sumatra
unpad	UNPAD	Universitas Padjadjaran — Jatinangor	Sumedang	West Java	Java
unri	UNRI	Universitas Riau	Pekanbaru	Riau	Sumatra
uns	UNS	Universitas Sebelas Maret	Surakarta	Central Java	Java
unsoed	UNSOED	Universitas Jenderal Soedirman	Banyumas	Central Java	Java
unsrat	UNSRAT	Universitas Sam Ratulangi	Manado	North Sulawesi	Sulawesi
unsri_indralaya	UNSRI Indralaya	Universitas Sriwijaya — Indralaya	Ogan Ilir	South Sumatra	Sumatra
untad	UNTAD	Universitas Tadulako	Palu	Central Sulawesi	Sulawesi
untan	UNTAN	Universitas Tanjungpura	Pontianak	West Kalimantan	Kalimantan
unud_jimbaran	UDAYANA Jimbaran	Universitas Udayana — Jimbaran	Badung	Bali	Bali
uny	UNY	Universitas Negeri Yogyakarta	Yogyakarta	DI Yogyakarta	Java
upi	UPI	Universitas Pendidikan Indonesia	Bandung	West Java	Java
usu	USU	Universitas Sumatera Utara	Medan	North Sumatra	Sumatra

The full list with additional metadata (area, compactness, centroid, CRS) is in reference/campus_registry.csv.

Directory structure

reference/     Campus registry — the join key for every other table (campus_id)
boundaries/    Campus boundary polygons (one .gpkg per campus)
grids/         Empty 20 m analysis grid per campus (geometry only)
cells/         Grid cells joined with SVI coverage/metadata (.csv + .gpkg), per campus
points/        Individual SVI image/pano locations (.gpkg), per campus
raw/           Raw scraped metadata from Google (raw/google/) and Mapillary (raw/mapillary/)
analysis/      Derived summary tables (analysis/tables/) and paper figures (analysis/figures/)

reference/campus_registry.csv

One row per campus: campus_id, display_name, full_name, city, province, island, centroid_lat/lon, area_km2, perimeter_km, utm_epsg (local UTM zone used for area/metric calculations), n_cells, cell_size_m (20 m), in_registry, verified.

boundaries/{campus_id}.gpkg

Campus boundary polygon (WGS84, EPSG:4326) used to clip the grid and define "inside campus" area.

grids/{campus_id}_grid.gpkg

The empty 20 m fishnet grid per campus before joining coverage data. Columns: grid_id, campus_id, row, col, cell_size_m, area_m2, area_inside_m2 (area of the cell that falls inside the campus boundary), frac_inside, bbox_west/south/east/north, cx/cy (centroid).

cells/{campus_id}_cells.csv / .gpkg and cells/all_campuses_cells.csv

The core analytical table: every grid cell joined with Mapillary (mly_*) and Google (ggl_*) coverage metrics. all_campuses_cells.csv stacks all 40 campuses (95,362 cells).

Key columns:

Column	Description
grid_id, campus_id, row, col	Cell identifiers
mly_count, ggl_count	Number of Mapillary images / Google panoramas in the cell
mly_n_sequences, mly_n_creators, mly_n_orgs	Mapillary contributor diversity
mly_year_min/max, ggl_year_min/max, mly_n_years, ggl_n_years	Temporal depth
mly_pano_ratio	Share of Mapillary images that are full panoramas (vs. perspective)
ggl_n_positions, ggl_n_historical, ggl_n_captures	Google position/historical-imagery counts
ggl_third_party_ratio, ggl_launch_ratio, ggl_scout_ratio, ggl_innerspace_ratio	Google capture-programme composition
ggl_mean_elevation	Mean camera elevation (m)
mly_coverage, ggl_coverage, either_coverage	Binary: cell has ≥1 image from that platform
agreement	both / mapillary_only / google_only / neither — spatial agreement between platforms
depth_diff	Difference in temporal depth (years) between platforms

points/{campus_id}_points.gpkg

Individual image/panorama locations, one GeoPackage per campus with two layers:

• mapillary: image_id, sequence_id, creator_id/creator_username, organization_id, captured_at/captured_date, compass_angle, altitude, camera_type, is_pano, raw + computed lon/lat.
• google: pano_id, tile_id, date/year/month, heading/pitch/roll, elevation, capture_source (launch / scout / third_party / innerspace), copyright_message, uploader, street_name, is_historical, parent_pano_id (links historical panoramas to their successor).

Both layers use CRS EPSG:4326.

raw/google/ and raw/mapillary/

Unprocessed per-campus scrape outputs, prior to grid aggregation:

• *_google_raw.csv — 406,808 rows total; full panorama-level metadata (position, heading/pitch/roll, capture programme, historical linkage).
• *_google_meta.csv — deduplicated one-row-per-panorama metadata.
• *_google_tiles.csv — raw tile-query results before deduplication/filtering.
• *_mapillary_raw.csv — 68,191 rows total; per-image Mapillary metadata (sequence, creator, camera type, capture time).
• *_*_progress.jsonl — scrape checkpoint/progress logs (retained for provenance/reproducibility, not needed for analysis).

analysis/tables/

Campus-level summary statistics used in the accompanying study:

• coverage.csv — coverage proportions per campus (mly_coverage, ggl_coverage, either_coverage, prop_both, prop_neither, mean depth difference).
• descriptives.csv — median images per covered cell, sequence/creator counts.
• contributors.csv / contributor_profile.csv — contributor concentration per campus (Gini coefficient, top-1/top-3 share, images needed for 90% of contributions).
• decay_profile.csv / decay_slope.csv — how coverage "openness" decays from campus core to edge.
• temporal_annual.csv / temporal_signature.csv — year-by-year and seasonal capture activity.
• geometry.csv — campus area, perimeter, compactness.
• programme.csv — Google capture-programme composition (launch/scout/innerspace/third-party) per campus.
• maup_wide.csv — sensitivity of coverage estimates to grid cell size (Modifiable Areal Unit Problem check).

analysis/figures/

PNG/PDF figures generated from the tables above (study area maps, coverage maps, agreement maps, temporal/decay/contributor plots, local Moran's I cluster maps).

Data sources & attribution

Imagery metadata was collected from two public street-level imagery platforms:

• Google Street View — panorama metadata retrieved via Google's Street View coverage/metadata endpoints.
• Mapillary — image metadata retrieved via the Mapillary API.

No image pixel data is included — only geolocated metadata (position, heading, timestamp, contributor/organization identifiers, capture-programme labels). Usage of this dataset must comply with the respective terms of service of Google Street View and Mapillary for any derived or redistributed use involving those platforms' data.

Coordinate reference systems

• All point/polygon geometries are stored in WGS84 (EPSG:4326).
• Area, perimeter, and grid-cell computations were performed in each campus's local UTM zone (see utm_epsg in reference/campus_registry.csv) before reprojection back to EPSG:4326.
• Grid cell size: 20 m.

Known limitations

• Coverage reflects platform-side data availability at the time of scraping, not ground-truth street network completeness.
• Mapillary coverage is strongly affected by volunteer/crowdsourced contribution patterns (see contributor concentration metrics — a small number of contributors dominate most campuses).
• Google coverage mixes multiple capture programmes (official Street View "launch", "scout" partner, "third-party", and "innerspace"/indoor) with different refresh cadences.
• is_historical Google panoramas represent superseded imagery kept for temporal-depth analysis, not current coverage.

Suggested citation

If you use this dataset, please cite the accompanying study (details to be added upon publication) and credit the underlying Google Street View and Mapillary contributors.

License

Released under CC BY 4.0 for the aggregated/derived metadata and analysis tables in this repository. This license does not extend to the original imagery or platform terms of Google Street View or Mapillary — see Data sources & attribution above.