Antalya Hotel Beaches: Measuring Sunbed Occupancy Without Identifying Guests

Define the metric first: occupancy is not footfall

Beach managers routinely conflate four different measurements: how many people walked onto the sand during the day (footfall), how many loungers are occupied right now (zone occupancy), how long a guest holds a lounger (dwell), and whether a queue is forming. Each question needs a different sensor and mounting geometry. A system that counts entries accurately can report occupancy poorly, and a system tuned for occupancy may miss short visits entirely.

Before choosing hardware, write down one sentence defining an 'occupied lounger'. The most practical definition: a lounger is occupied when a person is physically present in or beside its zone for a set window — say five to ten minutes. Track a second indicator separately: 'reserved but empty', where a towel and belongings sit without a body. That split is what solves the classic beach complaint of loungers bagged at six in the morning and left unused until the afternoon.

When occupancy is derived as entries minus exits, every small directional error compounds across the day. Pool staff, cleaning teams, deliveries and guests returning from the sea produce steady drift, and by late afternoon the number can be meaningfully wrong. The robust fixes are either to sense the whole zone directly from above the lounger rows, or to count directional entries and exits and reset the total on a known-empty condition — typically early morning before guests come out.

The seasonal context matters here. In peak summer months Antalya hotels ran occupancy near 84% in June, and the region welcomed over 17 million visitors across 2025. At that scale, beach zones genuinely overflow, and an accurate occupancy metric becomes an operational necessity rather than a nice-to-have.

• Footfall: how many people entered the beach in a period.
• Zone occupancy: how many loungers are held by bodies right now.
• Dwell time: how long a single lounger stays occupied.
• Towel reserve: a lounger held by belongings with no person present.

Anonymous sensing methods that never name a guest

The governing idea is detection rather than recognition: the system only needs to know that a person is present, never who it is. Overhead camera vision with person detection and tracking supplies direction, zones, dwell and headcount from a single stream. Track identifiers live only while someone is in frame; the frame is processed locally and only interval counts are retained.

Line crossing counts entries and exits at a choke point — for example, the stairway down to the sand or a walkway between lounger rows. Region or zone counting answers 'how many bodies are on this stretch of loungers right now'. Accuracy depends heavily on mounting: an overhead camera largely eliminates occlusion between people, while an oblique security-camera angle stacks guests behind each other and suppresses the count.

Where only occupancy is needed, time-of-flight depth sensors and thermal cameras are the simpler conversation. They produce no recognisable image, thermal units work in darkness, and a thermal sensor emitting only a count may not be processing personal data at all. The trade-off is context versus simplicity: a depth or thermal unit does one job extremely well and tells you nothing else, whereas vision adds queues, direction and dwell in exchange for careful handling of light and angle.

Wi-Fi and BLE counters look camera-free but count devices, not people: a phone left in the room, switched off or broadcasting a randomised MAC address vanishes from the count. At the same time, device identifiers count as personal data under GDPR in many regimes. For an open beach they are usually the weakest option on both accuracy and privacy.

Privacy and the law: Turkey and the European guest

In Turkey, processing images falls under Law No. 6698 on the protection of personal data, known as KVKK. The data protection authority consistently applies principles of data minimisation and proportionality: the narrowest useful camera angle, masking of unnecessary areas, a retention period limited to the stated purpose with automatic deletion, and a clear obligation to inform people under Article 10. Its guidance explicitly steers controllers away from more intrusive technologies such as facial recognition and voice recording.

Translated to a beach, this means concrete limits: cameras must not frame changing cabanas, showers or toilets, and filming guests in swimwear requires a legitimate purpose and a proportionality test. When the job can be done by an anonymous overhead count where faces are never resolved, both the legal conversation and the guest experience get easier. This is general information about the rules; for a specific deployment, consult a Turkish lawyer familiar with the Authority's practice.

Guests from Europe carry GDPR expectations with them. The key distinction is between detection (a person is present) and recognition (who the person is). Counting systems that never identify an individual and never link appearances across days or cameras sit comfortably in the anonymous video analytics category. Do not persist track IDs beyond the camera's field of view, do not reconcile a guest's visits across time, and aggregate figures over intervals rather than logging individual events.

• No facial recognition and no audio recording on the beach.
• Cameras kept away from cabanas, showers and toilets.
• Retention limited to purpose with automatic deletion.
• Guests informed under Article 10 of the KVKK.
• Counting is detection of presence, not identification of people.

Field accuracy, drift and the 'reserved but empty' lounger

Vendor accuracy figures are usually measured under conditions that do not resemble a beach. Real accuracy is cut down by groups walking abreast, by umbrellas and loungers that occlude the view, by oblique camera angles, and by backlit scenes at sunrise and sunset. Mount the camera overhead, then validate with a manual count at the busiest hour and worst lighting — that number, not the marketing figure, is the accuracy you actually have.

Drift is a separate failure mode. When the count is built on entries minus exits, a one or two percent error at every crossing accumulates into a meaningful discrepancy by the evening. The standard remedy is a periodic reset on a known-empty condition. For a beach, the natural reset point is early morning before the mass arrival, when most loungers are free.

The subtlest case is the 'towel without a body'. A thermal or vision sensor detects heat and motion, not a discarded towel, so it will honestly report a lounger as empty even though no free spot physically exists. That is not a sensor fault; it is a difference between metrics. Keep two numbers apart: real occupancy by bodies, and the share of loungers being 'held' by belongings. When the first is high and the second is climbing, that is a signal for staff to walk the row — but the decision to release or reassign a lounger belongs to a person, not to an algorithm.

Turning the occupancy number into staffing and capacity decisions

A metric earns its place when it is tied to an action. Set hour-by-hour thresholds: if zone occupancy stays above roughly 85–90% for more than forty minutes, increase lifeguards, bar staff and towel attendants for that zone. If the share of 'held but empty' loungers climbs toward midday, staff receive a verifiable task to walk the row and invite a release or reassignment.

Aggregating data by weekday and weather lets you see demand peaks and schedule staff more precisely than instinct. The useful property is that each observed event on the beach becomes a checkable task for a team member and a management metric, while the final call remains with a human. Start small: one zone, one sensor, a week of manual validation — then scale to the whole beach.

• An occupancy threshold triggers added staffing in the zone.
• A rising 'empty reserve' share creates a row-walk task for staff.
• Aggregation by day and weather sharpens shift planning.
• Run a single-zone pilot before scaling across the beach.