Retail Parking Lot Occupancy and Store Footfall Measurement

Measure real customer traffic at retail sites from space. Klarety processes high-frequency optical satellite passes to count vehicles in parking lots, compute hourly occupancy curves, and correlate traffic patterns with sales periods and competitor activity. Retailers benchmark site performance without deploying ground sensors.

Analyze footfall

Store traffic intelligence without ground sensor deployment

Klarety AI

Footfall benchmarks from satellite parking analysis

Klarety agents count vehicles from high-frequency optical passes and return hourly occupancy curves and competitor comparisons.

Optical

Vehicle counting and hourly occupancy curves

Agents classify parked vehicle density across lot polygons, compute occupancy rates, and correlate traffic with event calendars.

GIS Output

Occupancy layers for retail site GIS analysis

Export vehicle count grids and occupancy time-series for retail site selection model and trade area GIS tool integration.

Satellite vehicle counting for retail parking occupancy

Klarety agents detect vehicle presence in high-resolution optical imagery over defined parking lot polygons by identifying bright rectangular objects (vehicles) against asphalt. Vehicle count per polygon is normalized to lot capacity to yield occupancy rate. Weekly occupancy curves are benchmarked against competitors in the same trade area. Output is a weekly occupancy table, competitor comparison, and peak-traffic time classification.

Klarety AIMeasure parking occupancy at your retail sites

analysis/retail_parking_occupancy.pyAgent code

python

import eeimport numpy as npimport pandas as pd
ee.Initialize()
# Costco and competitor big-box stores, Sacramento areatarget_lot    = ee.Geometry.Rectangle([-121.38, 38.58, -121.36, 38.59])competitor_lot = ee.Geometry.Rectangle([-121.42, 38.57, -121.40, 38.58])
LOT_CAPACITY_TARGET    = 850  # estimated spacesLOT_CAPACITY_COMPETITOR = 720
def count_vehicles(aoi, start, end, capacity):    """Count bright rectangular objects on dark asphalt = vehicles."""    s2 = (ee.ImageCollection('COPERNICUS/S2_SR_HARMONIZED')          .filterBounds(aoi)          .filterDate(start, end)          .filter(ee.Filter.lt('CLOUDY_PIXEL_PERCENTAGE', 5))          .sort('system:time_start', False)          .first())
    b4 = s2.select('B4')    b8 = s2.select('B8')    b11 = s2.select('B11')
    # Vehicle: bright in visible, moderate NIR, low SWIR (not vegetation or metal)    vehicle_mask = b4.gt(2000).And(b8.lt(2500)).And(b11.lt(1500))
    # Morphological: vehicles are ~2-3 pixels at 10m, so connected components    vehicle_count_proxy = vehicle_mask.connectedPixelCount(100).gt(2)
    count = vehicle_count_proxy.reduceRegion(        ee.Reducer.sum(), aoi, 10, maxPixels=1e9    ).get('B4').getInfo()
    raw_count = int(count or 0)    # Calibration: each vehicle ~1.5 connected components at 10m    calibrated = int(raw_count / 1.5)    occupancy = min(calibrated / capacity * 100, 100)    return calibrated, round(occupancy, 1)
weeks = [    ('2026-03-01','2026-03-07','Wk1 (Mon)'),    ('2026-03-08','2026-03-14','Wk2 (Mon)'),    ('2026-03-15','2026-03-21','Wk3 (Mon)'),    ('2026-03-22','2026-03-28','Wk4 (Sat)'),]
print(f"{'Week':<14} {'Target Count':<14} {'Occ%':<8} {'Comp Count':<14} {'Comp Occ%'}")for s, e, label in weeks:    tc, to = count_vehicles(target_lot, s, e, LOT_CAPACITY_TARGET)    cc, co = count_vehicles(competitor_lot, s, e, LOT_CAPACITY_COMPETITOR)    print(f"{label:<14} {tc:<14} {to:<8.1f} {cc:<14} {co:.1f}")

output/retail_occupancy_report.mdOutput report