C-DRONE GUIDE · 21 AUGUST 2026
Parking occupancy count by drone: car park survey and price
A retail park car park packed on Saturday and empty on Tuesday, a company car park nobody can say for sure needs extending, a park-and-ride site whose real usage still has to be demonstrated before a development project : the occupancy-rate question keeps coming back, and neither manual counting nor ground sensors answer it well for a one-off or multi-site need. A drone hovering above the site, repeated at set intervals, photographs every space in a single pass and, once the vehicles are detected automatically, lets you rebuild a precise occupancy curve, space by space and hour by hour. Here is how a drone parking-occupancy mission works, and what it costs.
Published on 21 August 2026, reviewed on 21 August 2026 — regulations in force as of August 2026.
What a drone adds over manual counting or ground sensors
Manual counting requires one or more observers stationed at the entrances, forced to choose between tracking in/out flows or walking the aisles to log occupied spaces — a slow method, error-prone on large car parks, and impossible to repeat across several time slots in the same day without adding staff. Ground sensors (magnetic loops, per-space sensors) give a continuous reading, but need a fixed installation that is costly to deploy on an existing car park and rarely justified for a one-off study ahead of works.
A drone flips the problem : a vertical shot from 60-80 m up covers an entire car park of several hundred spaces in a single image, with no equipment installed on site. Processed by vehicle-detection software, each image becomes an exact space-by-space occupancy map. This approach builds on already-proven computer-vision methods : a study by Meng-Ru Hsieh, Yen-Liang Lin and Winston H. Hsu, presented in 2017 at the International Conference on Computer Vision and built on a dataset of nearly 90,000 vehicles photographed by drone across four car parks, shows that a detection network designed for aerial views clearly outperforms models built for ground-level images when it comes to counting parked vehicles (see the study on Google Scholar). Repeating the flight across several slots in a day, or over several representative days, turns that single photo into a genuine occupancy curve.
How a parking occupancy count mission works
Everything starts with choosing which time slots to observe, based on the question being asked : peak arrival time at an industrial site, Saturday afternoon for a shopping centre, market day for a municipal car park, or before/during/after a major event for a park-and-ride site. The remote pilot schedules a series of short hovering flights — a few minutes each — at those precise slots, keeping the same shooting position and altitude from one pass to the next so the images stay comparable. The method is close to the one used for trajectory reconstruction in traffic counting, but applied to parked rather than moving vehicles.
Each image is then processed by vehicle-recognition software that detects and locates every car, calculates the overall occupancy rate and, on car parks with marked bays, the rate per zone or row. The deliverable pairs an annotated map per slot, an occupancy curve over the period studied and a summary table that plugs directly into a development file or a presentation to a municipal council or management committee — image processing comparable to what we use for our photogrammetry missions.
Regulatory framework for hovering flight above a car park
A private car park (retail park, company site, healthcare facility car park) is overflown like any other professional site : agreement from the owner or manager, flight usually in sub-category A2 or A3 depending on the proximity of people moving between vehicles — neither drivers moving individually nor employees crossing the car park constitute an assembly in the regulatory sense, which allows overflight as long as safety distances are respected. A public car park falls under municipal roads : in built-up areas, professional open-category flight above public space has been allowed since 1 January 2026 (order of 23 December 2025), by day and without deliberately overflying people, but every pass still requires a prior declaration to the préfecture (cerfa 15476*04) with ten working days' notice — a single declaration can cover several planned slots over a given period.
The shooting altitude — 60 to 80 m is usually enough to cover a car park and detect every vehicle — stays well below the 120 m ceiling, which simplifies mission prep. As with any mission over an inhabited or commercial area, the remote pilot checks restricted or prohibited zones beforehand on the Géoportail drone map, especially near an aerodrome or a nearby sensitive site, and adapts the protocol if the car park hosts a large one-off gathering — a case covered in our guide on flying a drone at a festival or event.
The price of a drone parking occupancy study in 2026
The price depends on the number of slots photographed, the size of the car park and the depth of the analysis. Orders of magnitude observed in France in 2026 :
| Service | Observed price (excl. VAT) |
|---|---|
| One-off count, 3 to 5 slots on the same day with an occupancy map | €450 to €900 |
| Full study over several representative days (week + peak day) with an occupancy curve and report | €1,200 to €2,800 |
| Recurring follow-up (before/after development, several campaigns over the year) | +€300 to €600 per extra campaign |
This budget stays well below installing per-space sensors, which only makes sense for a car park meant to stay permanently instrumented, and settles a sizing question — extending, redesigning or, conversely, freeing up land — with factual data rather than a field impression. The simplest way to start a mission is to request a quote stating the slots to observe and the car park's size.
Frequently asked questions about drone parking counts
Can the drone tell reserved or disabled spaces apart? Yes, as long as the ground markings are visible in the image : the occupancy rate can then be broken down by space category, a useful figure for checking a public car park's compliance or the real use of reserved spaces.
How long does a counting session take? Each hovering flight lasts a few minutes ; a full day of spaced-out slots can be planned within a half-day on site, with the pilot running passes between processing steps.
Does the count work on an underground or covered car park? No — the drone needs a clear view from the sky. On a covered or underground car park, the alternative remains manual counting or per-space sensors ; the drone keeps its edge on open-air, surface-level or roof-deck car parks.
Can occupancy be compared before and after redevelopment? Yes, that is one of the most common uses : repeating the exact same viewpoint and time slots before and after reconfiguration or extension works measures the gain objectively, a useful argument in public procurement as much as in a management committee.