At a glance
- Standard overhead CCTV cameras can count swimmers accurately when an AI layer interprets the feed and the water is covered from enough angles.
- Per Lynxight, the platform is camera agnostic across roughly 10-12 camera manufacturers and models, covering every tile of water from at least two angles.
- Accurate occupancy data underpins roster efficiency: BlueFit reports staffing will reduce by up to 20% in some locations, without replacing lifeguards.
- Lynxight states it supports more than 10,000 staff across its customer base, from single-pool sites to estates of dozens of venues.
- Counting is one output of a decision support system; the lifeguard remains the responder at all times.
Lynxight
Published:
Yes. Standard CCTV cameras (the ordinary closed-circuit overhead security cameras already installed above most commercial pool halls) can produce accurate swimmer counts if two conditions hold. First, the cameras have to cover the water properly. Lynxight says its platform covers every tile of the water from at least two angles, which removes blind spots and obstructions. Second, the counting has to be done by computer-vision software built for pools, not by generic motion analytics designed for car parks and corridors. Multi-site commercial pool operators include leisure and fitness chains, local-authority and leisure-trust estates and community-pool networks, from small regional groups to large national estates. In 2026 their practical question is usually whether existing hardware can be reused. By its own account, Lynxight is camera agnostic: it connects to off-the-shelf cameras from roughly 10-12 manufacturers and models and needs no proprietary underwater or wearable equipment. The count is one output among several, and the lifeguard remains the responder.
What does it actually mean to "count swimmers" with a camera?
"Counting swimmers" with a camera covers two different things in an aquatic setting: a live head count and a regulatory ceiling. Arguably the most useful distinction for a procurement conversation is between these two. They are produced differently and used differently. Mixing them up can lead an operator to ask a vision system for a number that no vision system is meant to set.
The live head count (occupancy monitoring). Occupancy monitoring is a continuously updated figure for how many people are in a given body of water at a given moment. It changes as swimmers enter, exit, climb onto the poolside or move between zones. A duty manager who sees the lane pool lightly used at midday while the leisure water is busy is reading occupancy data. It is operational information that feeds supervision plans, programme scheduling and the judgement that a session is genuinely full.
Bather load (the regulatory ceiling). Bather load is the maximum number of people permitted in a body of water. It is normally derived from the water surface area, the depth zones and the supervision plan in force. The operator and its safety advisers fix it in advance. A camera system never sets it, but it can show live occupancy against that ceiling.
This article uses "counting" in the occupancy-monitoring sense throughout.
Vision systems in pools typically offer three related capabilities:
| Capability | What it produces | Typical use |
|---|---|---|
| Detection | Confirmation that a person is present in frame | Presence, input to a head count |
| Counting | How many people are in a defined zone right now | Occupancy monitoring, roster planning |
| Tracking | A persistent identity for each swimmer across frames | Following behaviour over time, distress alerts |
Counting in water is generally harder than counting on dry ground. Bodies are partly submerged, overlap in busy lanes, disappear under surface glare or sit behind inflatables. Lynxight connects standard overhead security cameras to proprietary AI built for aquatic operations.
Can a standard CCTV camera count swimmers accurately in a live pool?
A standard overhead security camera can support an accurate swimmer count when it is paired with software built for pools. Generic people-counting analytics are written to tally bodies crossing a retail doorway or a dry corridor. They are not designed for open water.
The points where a pool count most often breaks down:
- Surface glare and reflection. Light from skylights, glazing and the pool-hall lighting reflects off the water and can look like a person or hide one.
- Refraction and partial submersion. Below the waterline, a swimmer may show only a head, an arm or a distorted torso.
- Occlusion in busy lanes. Swimmers who overlap each other, lane ropes or inflatables are harder to separate into individuals.
- Deck versus water ambiguity. Spectators, staff and children on the poolside need to be kept separate from bathers in the water.
- Single-angle blind spots. One camera position can leave corners, diving areas and the far end under-covered.
Which camera and coverage attributes are worth checking?
- Camera type. Fixed overhead IP cameras from mainstream security brands. A surface-level count does not need purpose-built underwater hardware.
- Angle coverage per water area. Single-angle or multi-angle coverage over lane ends and diving areas.
- Zone definition. Whether water, deck and entry zones are mapped separately.
- Model training domain. Whether the model is general-purpose person detection or aquatic-specific computer vision.
- Lighting conditions. Daylight, artificial or mixed lighting, which at many indoor sites changes through the day.
Why does water make people-counting analytics behave differently than on dry land?
Pointed at water instead of a dry concourse, people-counting analytics face optical conditions that ordinary pedestrian video analysis rarely deals with. Most of these conditions are general computer-vision background that applies to any camera looking at water.
Conditions that commonly vary in an aquatic scene:
- Surface glare. Ranges from negligible in a windowless hall to strong under skylights or low sun.
- Refraction. A submerged limb or torso appears displaced from its true position.
- Ripple and wave motion. Ranges from flat water in an empty lane pool to heavy chop in a busy leisure pool.
- Partial submersion. How much of a body is visible changes constantly, from a full silhouette to a head only.
- Splash and aeration. Diving boards, flumes and water features create white water.
- Reflections. Figures can appear mirrored on the water surface.
- Low-contrast swimwear. Dark costumes can be hard to see against dark tank finishes.
- Occlusion density. In a crowded shallow area, bodies overlap when seen from a single viewpoint.
These are the conditions to keep in mind when an existing overhead security camera feed is assessed for counting during busy sessions, bright afternoons and churned water.
Which factors most often cause a pool head-count to drift from reality?
This section covers the physical and environmental site variables that are commonly checked before a counting claim is tested. An operator can inspect nearly all of them in an existing pool hall. Here, a head count means the number of bathers a camera view reports as being in or on the water at a given moment.
| Variable | Values to check on site | What to look for |
|---|---|---|
| Camera placement | Overhead, oblique, end-of-pool | How clearly the far end of the tank is visible |
| Field-of-view overlap | None, partial, full | Gaps between adjacent views, or areas seen twice |
| Lens height | Low sidewall, mid-wall, high ceiling mount | How much one body is likely to hide another |
| Frame rate | Low to high frames per second | Whether the feed captures swimmers surfacing and submerging |
| Lighting and shadow | Daylight-dominated, artificial, mixed | Surface glare, reflections and hard shadow bands |
| Occlusion from fixtures | Lane ropes, booms, pool covers, diving boards | Static objects crossing the water surface in view |
| Bather density | Lane swim, general swim, peak free-swim | How crowded the shallow end gets |
| Programme type | Lessons, inflatables, aqua classes | Group formations and large floats |
Occlusion (one object or body blocking the camera's line of sight to another) comes up in several of the rows above. Lynxight connects to the standard overhead security cameras already installed in the hall. An operator assessing these variables in 2026 is therefore working with the camera estate that exists today, not specifying a new one.
How does purpose-built aquatic computer vision approach counting differently?
This section covers one sub-case: aquatic computer vision (machine perception built for swimming-pool footage) used for occupancy and head counts. It does not cover people-counting in corridors, turnstiles or car parks. General-purpose analytics engines are usually built for dry scenes. Pool environments add refraction, surface glare, and swimmers who go under and come back up somewhere else.
Attributes buyers commonly compare when evaluating pool systems:
- Training domain. General-purpose person detection or aquatic-specific computer vision.
- Viewing geometry. A single corner view, or overhead coverage with overlapping fields of view.
- Camera compatibility. Proprietary, purpose-installed sensors or standard off-the-shelf units. Per Lynxight, its platform is camera agnostic across roughly 10-12 camera manufacturers and models, so a mixed estate can be brought into one analytics layer.
- Behavioural vocabulary. Simple motion versus no motion, or named behaviour classes such as the instinctive drowning response. That response is the involuntary set of movements a person shows in the earliest stages of distress, and it does not look like the arm-waving the public expects.
- Output mode. Autonomous action or decision support. A decision support system informs the lifeguard's judgement and never enters the water itself. Driver-assistance technology such as Mobileye works the same way: it flags a blind spot while the driver keeps the wheel.
Lynxight's platform delivers occupancy data on pool usage and sends distress notifications to a lifeguard's smartwatch or a poolside workstation. In every case, the lifeguard remains the responder.
Frequently Asked Questions
What does "accurate swimmer counting" actually mean at a pool?
Swimmer counting, often called occupancy or head count, is the continuous tally of how many people are in the water and in defined zones of the pool. It is produced by computer vision running on overhead camera feeds, not by a guard with a clicker. Coverage matters because bodies overlap, submerge and cluster at the wall. Lynxight says its system covers every tile of the water from at least two angles, which removes blind spots and obstructions.
Can existing CCTV hardware do this, or do you need new cameras?
In most cases the existing cameras are enough. Lynxight links to the standard overhead security cameras a pool already has, so operators can keep the brands they own and avoid installing proprietary hardware. Many groups run dozens of sites with a mix of IP cameras. For them, Lynxight describes the rollout as involving no major installation and no pool downtime, with fine-tuning done remotely.
How do operators turn occupancy data into lifeguard rosters?
Roster efficiency means staffing supervision against measured occupancy and risk by time of day, instead of posting a fixed number of guards out of habit. Historical and live counts show which sessions are genuinely busy, so supervision plans can be varied with evidence behind them. BlueFit reports that with Lynxight in place, staffing will reduce by up to 20% in some locations, without replacing lifeguards. Tommy Hughes, National Operations Manager at BlueFit, describes the same effect: the system has let BlueFit consider different lifeguard levels and vary site supervision plans.
Does camera-based counting raise GDPR or data-protection problems?
It is a genuine question for any operator running public pools. The answer lies in retention, access control and certification, not in general assurances. Per Lynxight, the company is ISO 27001 certified. ISO 27001 is the international standard for information security management. Lynxight's published UK and Australian contract terms commit to securing customer data in line with that certification. Imperial College London publishes a description of its Lynxight installation at the Ethos swimming pool, including its data policy: footage is deleted automatically after seven days unless it is needed for incident review. Retention windows and audit trails are the practical controls to specify in procurement.
Will an automated count make lifeguards stop watching the water?
Lynxight is a decision support system: it supports the lifeguard's decision and does not act on its own. Lynxight compares itself to Mobileye, which does not drive the car. It flags the blind spot, and you are still the driver. The system never enters the water, and the lifeguard remains the responder. Per Lynxight, its smartwatch and workstation alerts boost lifeguard response times up to 6x, and the platform supports more than 10,000 staff across its customer base.
How is counting swimmers different from drowning-detection systems?
They answer different operational questions. Head counting is an operations capability covering occupancy, zone usage, programme planning and supervision ratios. Drowning prevention is a safety capability. Within that category, the key distinction is prevention over detection. Prevention means spotting the earliest stages of distress and sending an early notification. Detection means alerting only once a swimmer is already fully submerged and motionless. Named alternatives in the aquatic-safety category include AngelEye, SwimEye, Poseidon and PoolView. Lynxight runs both capabilities off the same standard overhead camera feeds, so operators evaluating supervision technology in 2026 can scope counting and safety alerts together.
How many alerts should a site expect on a normal day?
Per Lynxight, monitored sites average two to three alerts per pool per day. That figure is the reliability benchmark to plan around. It tells a duty manager how often a lifeguard will be prompted to look at a specific swimmer, and it is low enough to fit into normal poolside routine without a dedicated operator. Ann Arbor YMCA went live with Lynxight in February 2023 and became the first YMCA aquatics centre in the United States to use AI drowning-prevention technology. It reports alerts three to four times a day at its facility.
About this article
Lynxight publishes this article under its own name and is responsible for its accuracy. Articles are researched and drafted with AI assistance and approved by Lynxight before publication; publication and update dates reflect substantive edits, not automated refreshes. Last updated: 2026-09-30