Missing Women: Measuring Gender Representation in Public Spaces

Analyzing gender representation in Mumbai and Navi Mumbai using GoPro wearable camera footage.

Key Findings

Summary

City	Images	Pedestrians	Prop. Female	Sex Ratio
Mumbai	2,251	12,382	19.3%	239
Navi Mumbai	1,239	4,951	18.2%	222

Women are significantly underrepresented in public spaces, comprising only 18-19% of pedestrians. The pedestrian sex ratio (females per 1000 males) is far below census baselines (Mumbai: 838, Navi Mumbai: 910).

By Mode

Mode	Mumbai	Navi Mumbai
Pedestrians	19.3%	18.2%
Two-wheelers	8.4%	5.7%

Women are far less represented among two-wheeler riders than pedestrians.

By Road Type

City	Primary	Secondary	Tertiary	Residential
Mumbai	14.7%	14.1%	16.9%	17.1%
Navi Mumbai	17.9%	15.5%	16.4%	14.9%

Pipeline

Raw Videos → EXIF Extraction → GPS Parsing → GPS Index →
Annotations → GPS Assignment → Geo Enrichment → Analysis Data →
EDA + Analysis + Maps

Run Full Pipeline

# Run E2E for all cities
python scripts/run_pipeline.py --city all --skip-osm

# Run for single city
python scripts/run_pipeline.py --city mumbai --skip-osm

# Skip visualization steps
python scripts/run_pipeline.py --city all --skip-osm --skip-viz

Pipeline Scripts

Script	Description
00_extract_exif.py	Extract EXIF metadata from video files
01_parse_exif.py	Parse EXIF text to GPS timeseries CSV
04_build_gps_index.py	Build GPS lookup index (parquet)
05_parse_annotations.py	Parse Label Studio JSON exports
06_assign_frame_gps.py	Match frames to GPS coordinates
07_enrich_with_geo.py	Add road type from itineraries
08_build_analysis_data.py	Build final analysis dataset
09_eda.py	Exploratory data analysis plots
10_analysis.py	Publication tables and figures
11_make_maps.py	Interactive and static maps
run_pipeline.py	Run full E2E pipeline

Setup

# System tools (macOS)
brew install ffmpeg exiftool

# Python environment
pip install pandas numpy geopandas folium contextily statsmodels scikit-learn pyyaml tqdm

Directory Structure

data/
  {city}/
    exif/                    Raw EXIF text files
    exif_metadata/           Parsed GPS timeseries CSVs
    gps_index/               GPS lookup parquet files
    labelstudio/             Label Studio JSON exports
    analysis_data.parquet    Final analysis dataset
sampling/
  {city}/
    itineraries/             Road type classifications
figs/                        Generated figures and maps
tabs/                        Generated LaTeX tables
scripts/                     Pipeline scripts
cities.yaml                  City configuration

Outputs

Data

• data/{city}/analysis_data.parquet - Primary analysis dataset
• data/{city}/analysis_data_long.parquet - Long format (all annotators)

Figures (figs/)

• fig2_distribution.pdf - Proportion female distribution
• fig3_multipanel.pdf - Multi-panel summary
• fig4_weekday_weekend.pdf - Weekday vs weekend comparison
• fig5_pedestrian_loess.pdf - LOESS: pedestrian women vs crowd size
• map_locations.pdf/html - Data collection locations
• map_sex_ratio.pdf/html - Sex ratio by location
• eda_*.pdf - Exploratory analysis plots

Tables (tabs/)

• table1_city_summary.tex - City-level summary
• tableS1_road_type.tex - By road type
• tableS2_temporal.tex - Temporal patterns
• tableS3_poi_infrastructure.tex - POI and infrastructure

Data Coverage

Metric	Mumbai	Navi Mumbai
GPS coverage	97.4%	99.4%
Itinerary match	66.3%	72.1%
Hour range (IST)	7-19	9-18

License

MIT