Parking Space Detection on Unmarked Lots

Parking space detection on unmarked gravel, grass and event lots

Operators of rural car parks, trailheads, seasonal overflow fields and event parking often ask a reasonable question: can parking space detection work where there are no painted bays? The practical answer is yes, but with an important condition that must be made clear from the start.

Where there are no marked spaces, there is no such thing as per-space occupancy. A system cannot honestly say that space 18 is free if space 18 does not exist on the ground. On an unmarked lot, Parkinto reports a count of parked vehicles and how full the defined area is. It does not report the free or occupied state of a numbered bay.

That distinction matters. Count-based data is often exactly what a municipality, rural site or event operator needs: how many cars are present, whether the area is approaching its practical limit, and whether visitors should be sent elsewhere. But it is not the same as individual bay monitoring, and stakeholders should not be allowed to assume that every vehicle position will be tracked as a separate parking space.

What count-based data can answer, and what it cannot

On an unmarked lot, the useful operational unit is the area, not the bay. That area may be a gravel yard, a grass field, a trailhead car park, a village parking area or a temporary event field. The system counts vehicles visible in that area and compares the count with an operator-defined capacity or fullness threshold.

This is the central operating change: the data supports capacity management rather than bay-by-bay allocation. For many outdoor sites, that is the right level of control. A trailhead car park usually needs to know whether visitors should continue driving toward the location or choose another site. A seasonal field needs to know when stewards should open the next section. A village needs to know when parking pressure is building around popular access points.

Where parking space detection needs marked bays

For marked outdoor lots, Parkinto can report the free/occupied state per space in current operation at 99.7% accuracy from a camera image captured and evaluated every 30 seconds, with the camera positioned so the occupancy is visible to the naked eye from that viewpoint; one well-placed camera covers up to 100 spaces. An independent measurement by the Czech Technical University put an earlier version at 99.4%.

Those figures apply to per-space occupancy where spaces are visible and defined. They should not be transferred to an unmarked grass field as if the system were identifying numbered spaces that are not there. If the operator needs to guarantee that a particular disabled bay, permit space, EV charging space or reserved visitor space is free or occupied, that space should be physically marked and treated as an individual monitored space.

In practice, many sites use both ideas. A small set of important marked spaces can be monitored individually, while the surrounding unmarked area is managed by count. The key is to describe each area correctly: per-space occupancy where spaces exist, vehicle counts where they do not.

Why ground-based approaches struggle on gravel, grass and temporary surfaces

Unmarked outdoor parking areas are often chosen because they are flexible and inexpensive. That same flexibility makes ground-based detection impractical. On a gravel yard, vehicle positions shift over time, stones move, potholes appear and the surface may be regraded. On grass, the usable parking pattern changes with mowing, mud, wet ground, snow cover and recovery periods. On temporary event land, the parking layout may exist for a few days and then disappear.

Installing hardware in or on the ground assumes that each parking position is stable. Unmarked lots break that assumption. If drivers do not align to fixed bays, a sensor assigned to a theoretical spot may sit under the wrong part of a vehicle, between vehicles, or outside the area actually used that day. Temporary mats, cones and stewarded rows can improve order, but they still do not create permanent, numbered spaces unless the operator deliberately lays them out and maintains them.

This is why counting from a visible viewpoint is often more realistic. The operator defines the area to monitor, keeps the camera view clear, and uses the count to manage capacity. The system follows what is actually happening on the surface rather than forcing an artificial bay map onto a site that does not behave like one.

Whole-area counting or virtual zones?

When planning parking space detection for an unmarked lot, the first design choice is whether to count the whole area as one unit or split it into virtual zones. A whole-area count is simplest. It works well when the operator only needs one answer: is this site available, limited or full?

Virtual zones are useful when different parts of the site are operated differently. A rural municipality may want separate counts for the main gravel area and a grass overflow field. An event operator may want to know when the first section is close to full before opening the next. A trailhead may have an upper area and a lower area that visitors experience as separate choices.

The rule is practical: define zones only where the information will change an operational decision. If nobody will act differently when Zone A is full and Zone B is not, a single count may be better. If signage, staff instructions or visitor information will change by area, zones are useful.

• Use one area when visitors see the car park as a single destination and the operator only needs an overall status.
• Use virtual zones when overflow, event sections, access roads or visitor routing are managed separately.
• Avoid over-zoning where there are no physical or operational boundaries. Too many zones can imply a precision that the site layout does not support.
• Review the capacity threshold with local staff. The practical full point may be lower than the maximum number of vehicles that can physically fit.

What changes across seasons on an overflow field

Seasonal and overflow parking is rarely static. In dry summer weather, a grass field may accept orderly rows and a higher practical capacity. After heavy rain, soft ground may reduce the usable area. In winter, snow banks, reduced visibility and changed access routes can alter where drivers actually park. During peak hiking, cycling or event days, drivers may park more tightly than on quiet days; when stewards are present, rows may be more consistent.

For operators, the important point is that the count should be interpreted against the current operating plan. A field may have one practical capacity in dry conditions, another when only a section is usable, and another when the operator closes part of it for ground recovery. In 2026, the most useful setup is not a fixed theoretical bay plan for an unmarked surface. It is a clear count, paired with seasonal rules for how full is full.

Before each season or event period, confirm three things: the monitored boundary, any active virtual zones, and the fullness thresholds used for public messaging. If the usable field changes, update the operational interpretation rather than pretending the site still contains the same number of imaginary spaces.

Real example: Krásná in the Karlovy Vary region

A published Parkinto example shows how this works in practice without marked bays. In the mountain village of Krásná in the Karlovy Vary region, Parkinto works with Czech partner ANTONYapp from Otrokovice. The system covers the two main parking areas, Zlatník and Visalaje, which are used by hikers and cyclists.

These parking areas have no marked spots, so the system tracks the number of parked cars rather than the state of individual spaces. The village shares availability with visitors. That is the correct expectation for an unmarked rural setting: count the cars, communicate availability, and help visitors make better decisions before they add pressure to the site.

How to set expectations with stakeholders

The hardest part of parking space detection on unmarked lots is often not technical. It is language. Stakeholders may hear the word detection and assume that every vehicle position becomes an individual space. Operators should correct that early, especially when elected officials, tourism teams, event organisers or local residents expect a live map of free spaces.

A clear explanation usually helps: if there are no painted bays, there are no numbered spaces to report. The system can still say how many vehicles are present, whether the site is filling up, and whether to show availability or full status to visitors. It cannot truthfully say that a specific unpainted place is free.

• Use the phrase vehicle count for unmarked areas.
• Use the phrase per-space occupancy only where spaces are physically marked.
• Agree public messages in advance, such as available, limited spaces, use overflow, or full.
• Document the threshold for full so staff, signs and websites use the same logic.
• Mark bays where individual guarantees matter, such as reserved, accessible or charging spaces.

For rural, municipal, seasonal and event parking, this honest distinction prevents disappointment and makes the data more useful. The goal is not to pretend a field is a lined car park. The goal is to give operators a reliable operational view of parking pressure and give visitors a better chance of choosing the right place to go.

The Krásná deployment is written up in full in our note on the installation on an unmarked parking lot.

Faded or missing markings are one of several site conditions worth testing before you buy — our field guide to parking space detection failure modes covers the rest.

If you manage an outdoor unmarked lot and need practical visibility of vehicle counts, fullness and visitor availability, talk to Parkinto about the right setup for your site.