AI Drowning Detection & Aquatic Safety Monitoring System (Overhead + Underwater Vision, Lifeguard Smartwatch Alerts)

Grid-powered AI drowning detection system for public pools and water parks: overhead 4K + underwater cameras, edge AI swimmer tracking, sub-10-second alarms to lifeguard smartwatches, strobes and PA — engineered end-to-end by REDCOAST.LTD.

All Products
Model RC-DDS-600
ai-video-analyticsdrowning-detectionpool-safetyedge-aismart-citypublic-safetyiot

Overview

The RC-DDS-600 is a grid-powered, AI-driven drowning detection and aquatic safety monitoring system engineered for public swimming pools, water parks, hotel and resort pools, school and university aquatic centers, and swim training facilities. It combines overhead 4K cameras, optional underwater cameras, and an on-site edge AI unit to continuously track every swimmer in the water, recognize submersion and distress events in real time, and push alarms to lifeguard smartwatches, poolside strobe beacons, and the facility PA system within seconds. Even well-trained lifeguards face scanning fatigue, surface glare, refraction, and crowded water — studies consistently show that a submerged victim can go unnoticed for critical minutes in a busy pool. The RC-DDS-600 closes that vigilance gap: it never blinks, it watches below the surface where human eyes cannot, and it turns every second saved into survival probability. REDCOAST.LTD delivers it as a complete solution — cameras, self-developed electronics, edge AI, web management platform, and mobile app — designed, built, and integrated by one team.

Key Features

  • Dual-view perception: overhead 4K cameras track surface behavior and swimmer trajectories; IP68 underwater cameras see submerged swimmers that lifeguards physically cannot, eliminating blind spots caused by glare and refraction.
  • Edge AI event engine aligned with ISO 20380 detection logic: motionless-swimmer detection, submersion beyond a configurable time threshold (10–30 s), and abnormal distress-motion patterns, with self-learning false-alarm suppression per pool zone.
  • Sub-10-second alerting chain: alarms reach lifeguard smartwatches, poolside strobes/sounders, the control-room workstation, and the facility PA over the local network — no internet dependency for life-safety alerts.
  • Continuous swimmer tracking of up to 250 concurrent bathers with per-zone occupancy counting, lane usage statistics, and heatmaps for staffing and operations planning.
  • Self-developed hardware: REDCOAST.LTD designs its own multi-channel PoE++ camera interface board, edge AI inference carrier board, underwater camera signal-conditioning PCB, supervised alarm I/O board, and LiFePO4 UPS power-management PCB — every board can be adapted to the project.
  • Retrofit friendly: ONVIF-compliant, so existing overhead CCTV cameras can be reused in overhead-only configurations to reduce cost for smaller pools.
  • Privacy by design: all video analytics run on the local edge unit; optional automatic face blurring and configurable retention support strict data-protection regimes worldwide.
  • After-hours protection mode: outside opening hours the same cameras detect unauthorized entry into the water and trigger intrusion alarms — one system, two safety functions.
  • Full management stack included: web platform for incident review, audit logs, and multi-site fleets, plus a mobile app for managers and duty supervisors.

Technical Architecture

The system is organized in three layers. The perception layer consists of ceiling- or pole-mounted overhead 4K cameras (one camera typically covers up to ~300 m² of water surface depending on mounting height and geometry) and wall-recessed or surface-mounted underwater cameras in 316L stainless-steel housings with optically corrected acrylic windows. Both camera families terminate on REDCOAST.LTD's self-developed multi-channel PoE++ interface board, which provides per-port power budgeting, surge protection, and link diagnostics, and on a dedicated underwater camera signal-conditioning PCB that handles long cable runs from pool walls to the equipment room.

The edge intelligence layer is the RC-DDS-600 edge AI unit: a fanless industrial controller built around a REDCOAST.LTD carrier board with an NPU delivering up to 32 TOPS. It decodes up to 32 camera streams, runs person detection, pose estimation, and multi-object trajectory tracking, then fuses overhead and underwater views into a unified 3D model of each swimmer's position and depth. The event engine classifies situations — a swimmer motionless at the surface, a body descending and remaining below the surface past the configured threshold, erratic vertical distress motion — and scores them against zone-specific baselines that the system learns over the first weeks of operation (diving wells, slide splash-down zones, and teaching areas each behave differently).

The alerting and management layer dispatches confirmed events in parallel: over the local Wi-Fi 6 network to rugged lifeguard smartwatches showing the alarm type and pool-map location; through the supervised alarm I/O board to strobes, sounders, and the facility PA/voice-alarm system; and to the control-room workstation with live video pop-up. Every event is logged with a video clip for debriefing and liability documentation. A cloud connection (Ethernet, fiber, or 4G/5G backup) synchronizes logs, analytics, and multi-site dashboards to the REDCOAST.LTD web platform — but detection and alarming remain fully functional if the facility loses internet connectivity.

Connectivity & Power

The RC-DDS-600 is a mains-powered (grid) product: swimming facilities are by definition powered buildings or serviced outdoor venues, so no solar configuration applies. The edge AI unit and PoE++ camera power system accept AC 100–240 V, 50/60 Hz, with total consumption from roughly 60 W (small 8-channel retrofit) to 450 W (large 32-channel water park deployment including camera PoE budget). An integrated LiFePO4 UPS module rides through power dips and keeps detection and alarming alive for 30–60 minutes during an outage — precisely the situation in which supervision must not fail. Networking options include Gigabit Ethernet, 10G SFP+ fiber uplinks for multi-pool campuses, a dedicated Wi-Fi 6 access-point channel for the lifeguard smartwatches, and an optional 4G/5G router for cloud backhaul at venues without fixed broadband. ONVIF, RTSP, REST API, and MQTT interfaces allow integration with existing VMS, BMS, and facility PA systems.

Protection & Reliability

Indoor pool halls are among the harshest electronic environments in civil buildings: 100 % condensing humidity, chloramine-laden air, and washdown cleaning. Overhead cameras are rated IP66 with anti-fog heated front glass and corrosion-resistant coated housings; underwater cameras are IP68, pressure-rated to 1.5 bar, built in 316L stainless steel, and sealed for continuous immersion in chlorinated or saltwater pools. Outdoor variants operate from -30 to +60 °C with strong-sunlight HDR imaging for glare-heavy open-air pools. The edge unit is fanless, uses industrial-grade components (0 to +50 °C equipment-room rating), mirrors recordings across dual storage drives, and supervises itself with hardware watchdogs, camera-tamper detection, and line-monitoring on all alarm outputs, so a cut cable or failed strobe is reported as a fault rather than discovered during an emergency. The design targets a service life of 10 years with proactive health monitoring from the cloud platform.

Application Scenarios

  • Municipal and community indoor pools: augment small lifeguard teams during crowded public sessions and school hours; the system watches the bottom of every zone while staff manage the deck, cutting detection time for submersion events from minutes to seconds.
  • Water parks and wave pools: agitated, opaque water and huge bather loads overwhelm human scanning; per-zone AI baselines handle wave cycles and slide splash-downs while smartwatch alerts route each alarm to the nearest guard station.
  • Hotel and resort pools: many resort pools are unguarded or lightly guarded; RC-DDS-600 provides continuous supervision, after-hours intrusion detection, and documented duty-of-care that insurers increasingly ask for.
  • School and university aquatic centers: protect learn-to-swim classes and training sessions, with lane-usage analytics that help schedule coaching and public access.
  • Competition and training venues: monitor breath-hold training risks (shallow-water blackout) in diving wells and 50 m pools where depth makes visual supervision hardest.
  • Residential community pools: overhead-only configurations give property managers an affordable safety layer plus occupancy analytics for maintenance planning.

Case-style Examples

  • Municipal aquatic center retrofit: a 25 m, 6-lane public pool with an existing ONVIF CCTV system needed drowning detection without rebuilding the hall. REDCOAST.LTD reused four existing overhead cameras, added six wall-mounted IP68 underwater cameras during a scheduled drain-down, and installed a 16-channel edge unit in the plant room. Lifeguards received smartwatch alerts from day one; in the first season the system flagged two genuine submersion incidents in under 15 seconds, both resolved without injury.
  • Water park multi-pool deployment: a large water park with a wave pool, kids' lagoon, and slide splash pools deployed a 32-channel system with zone-specific AI baselines and three guard-station strobe/sounder loops driven by the supervised I/O board. Heatmap analytics also revealed overcrowding patterns that reshaped the park's guard rotation and admission pacing.
  • Resort outdoor pool with heavy glare: an oceanfront resort pool suffered constant surface glare that made both human and camera surface-observation unreliable. The configuration prioritized underwater cameras with HDR overhead units on powder-coated corrosion-resistant poles; saltwater-rated 316L housings handled the chlorinated seawater fill, and the unguarded night period is covered by after-hours intrusion mode pushing alerts to the duty manager's app.

Customization & Selection Guide

Choose the camera mix first: overhead-only (budget retrofits, shallow residential pools), overhead + underwater (recommended for any pool deeper than 1.4 m or with public sessions), or underwater-heavy (glare-prone outdoor and deep diving pools). Size the edge unit by channel count: 8 channels for a single small pool, 16 for a standard municipal center, 32 for water parks or multi-pool campuses — units can be cascaded for larger estates. Select alert endpoints to match staffing: smartwatches and strobes for guarded pools; app push, PA voice messages, and remote monitoring-center forwarding for unguarded ones. Because REDCOAST.LTD designs its own PCBs, project-level adaptations — extra supervised outputs for legacy alarm panels, custom housing geometries for architectural ceilings, saltwater or thermal-spa variants — are engineering options, not exceptions.

Deployment & After-sales

Underwater cameras install in pool walls during a scheduled drain-down or via surface-mount kits that avoid draining entirely; overhead cameras mount to ceilings, trusses, or powder-coated poles for outdoor pools. A typical single-pool deployment completes in 3–5 working days including calibration, followed by a 2–4 week self-learning period during which sensitivity is tuned against real bather behavior. REDCOAST.LTD provides remote commissioning support, lifeguard and operator training materials, firmware and AI-model updates over the air, spare-parts availability, and SLA-based technical support with remote diagnostics through the cloud platform.

Standards & Compliance

The detection logic and alerting chain are designed toward ISO 20380:2017 (computer-vision drowning-detection systems for public pools) and the German DIN 79002 direction, complementing facility safety management per EN 15288. Electronics follow CE/RoHS and IEC 62368-1; ingress protection is rated per IEC 60529 (IP66 overhead / IP68 underwater); cameras are ONVIF conformant. Privacy-relevant processing stays on the edge device with optional face blurring and configurable retention, supporting GDPR-style data-protection requirements in any jurisdiction. The system is designed for worldwide deployment across climates, water chemistries, and regulatory environments.

Why REDCOAST.LTD

REDCOAST.LTD is an end-to-end smart-IoT solution provider: we deliver the complete system — cameras, edge AI hardware, alarm devices, web platform, and mobile app — integrated by one engineering team. What sets us apart from integrators assembling off-the-shelf parts is that we develop our own hardware, down to custom PCBs: the PoE++ camera interface, underwater signal-conditioning, edge AI carrier, supervised alarm I/O, and UPS power-management boards in the RC-DDS-600 are all REDCOAST.LTD designs. That means every layer can be adapted to your pool geometry, alarm infrastructure, and regulatory context — and supported for the long life of the installation.

Contact REDCOAST.LTD to discuss a customized drowning-detection configuration for your pool, aquatic center, or water park.

Specifications

AI Detection & Analytics

Detection Events
Submersion, motionless swimmer, distress motion, after-hours intrusion
Submersion Alert Threshold
10-30 (configurable) s
Alarm Dispatch Latency
<3 (from event confirmation) s
Concurrent Swimmer Tracking
up to 250 persons
Coverage per Overhead Camera
up to 300 m²
Analytics
Occupancy count, zone heatmap, lane usage, incident video clips
Self-learning Calibration Period
2-4 weeks

Camera Subsystem

Overhead Camera Resolution
3840×2160 (8 MP)
Overhead Camera Dynamic Range
120 (HDR, glare suppression) dB
Underwater Camera Resolution
1920×1080 / 2560×1440
Underwater Housing Material
316L stainless steel, acrylic optical window
Underwater Pressure Rating
1.5 bar
Camera Power
PoE IEEE 802.3at/bt
Interface Standard
ONVIF Profile S/T, RTSP

Edge AI Unit

AI Compute
up to 32 TOPS
Camera Channels
8 / 16 / 32 ch
Storage
2 × 4 (mirrored) TB
Video Retention
7-90 (configurable) days
Cooling
Fanless, industrial-grade
Reliability Functions
Hardware watchdog, camera tamper detection, supervised alarm lines

Alerting

Lifeguard Smartwatch
Wi-Fi 6 / BLE 5.0, IP68, vibration + pool-map alarm display
Watches per System
up to 32
Local Alarm Outputs
8-16 supervised relay/0-24 V channels (strobe, sounder, PA trigger)
PA / Voice Alarm Integration
Dry contact, SIP, or line audio
Workstation & App
Control-room client with live pop-up, iOS/Android manager app
Offline Operation
Full detection and local alarming without internet

Power (Grid)

Input Voltage
AC 100-240, 50/60 Hz V
System Power Consumption
60-450 (incl. camera PoE budget) W
PoE Budget
up to 720 W
UPS Backup
LiFePO4, 30-60 min autonomy
Surge Protection
Per-port PoE surge protection + AC SPD Type 2/3

Environment & Protection

Overhead Camera IP Rating
IP66
Underwater Camera IP Rating
IP68 (continuous immersion)
Operating Temperature (Outdoor Cameras)
-30 to +60 °C
Operating Temperature (Edge Unit)
0 to +50 °C
Humidity Tolerance
up to 100 (condensing, pool-hall rated) %RH
Water Chemistry Compatibility
Chlorinated, saltwater, thermal-spa variants
Anti-fog Front Glass Heating
Automatic, thermostat-controlled

Connectivity

LAN
Gigabit Ethernet, 10G SFP+ fiber uplink
Smartwatch Network
Dedicated Wi-Fi 6 AP channel
Cloud Backhaul
Ethernet / fiber / optional 4G/5G router
Integration APIs
REST, MQTT, ONVIF, SIP
Platform
REDCOAST.LTD web platform + mobile app, multi-site dashboard

Capabilities — configurable per project

Specifications are tailored to each project — the options below show what we can support.

Camera Configuration

  • Overhead-only (retrofit existing ONVIF CCTV)
  • Overhead + underwater (recommended)
  • Underwater-heavy (deep / glare-prone pools)

System Scale

  • 8-channel single pool
  • 16-channel aquatic center
  • 32-channel water park / multi-pool campus
  • Cascaded multi-unit estate

Alert Endpoints

  • Lifeguard smartwatches + strobes
  • PA / voice-alarm integration
  • Manager app + monitoring-center forwarding
  • Legacy alarm-panel supervised outputs

Pool Type Variant

  • Indoor chlorinated
  • Outdoor high-glare
  • Saltwater / seawater fill
  • Thermal spa / high-temperature

Privacy Options

  • On-edge processing only
  • Automatic face blurring
  • Configurable retention 7-90 days

Related solution guidance

Frequently Asked Questions

How does an AI drowning detection system work?

Overhead and underwater cameras stream video to an on-site edge AI unit that tracks every swimmer's position and depth in real time. When it detects a swimmer submerged beyond a set threshold (typically 10-30 seconds), motionless at the surface, or showing distress motion, it pushes an alarm with the pool-map location to lifeguard smartwatches, strobes, and the PA system within seconds. All analysis runs locally, so alarms work even without internet.

Does the RC-DDS-600 replace lifeguards?

No. It is a lifeguard-assistance and supervision-augmentation system, consistent with ISO 20380 positioning. It compensates for human limits — glare, refraction, scanning fatigue, crowded water — and shortens detection time, but trained staff still perform the rescue. At unguarded pools it provides a documented safety layer and routes alerts to duty managers or a monitoring center.

Can it work with our existing pool CCTV cameras?

Often yes for the overhead layer. The system is ONVIF-compliant, so existing overhead IP cameras with adequate resolution and viewing angles can be reused in retrofit configurations. Underwater coverage requires the dedicated IP68 underwater cameras, which can be installed during a scheduled drain-down or with surface-mount kits that avoid draining the pool.

How fast does the system detect a drowning incident?

The submersion threshold is configurable from 10 to 30 seconds depending on pool type and zone; once an event is confirmed, alarm dispatch to smartwatches and strobes takes under 3 seconds. In practice the total time from a swimmer going under to a lifeguard being alerted is well under 30 seconds — versus the several minutes a submerged victim can go unnoticed in a crowded pool.

Does it handle outdoor pools with strong sunlight and glare?

Yes. Outdoor configurations use 120 dB HDR overhead cameras with glare suppression and lean more heavily on underwater cameras, which are unaffected by surface glare. Outdoor camera variants operate from -30 to +60 °C and are available in saltwater-rated 316L housings for seawater-fill and coastal pools.

What about swimmer privacy and GDPR-type regulations?

All video analytics run on the local edge unit — footage does not need to leave the facility. Optional automatic face blurring, configurable retention (7-90 days), role-based access, and audit logs support GDPR-style compliance. The cloud platform receives only event metadata and health telemetry unless the operator explicitly enables clip sync.

What standards apply to drowning detection systems?

The key standard is ISO 20380:2017, which defines safety requirements and test methods for computer-vision drowning-detection systems in public pools; Germany's DIN 79002 is a related national benchmark. The RC-DDS-600 detection logic and alerting chain are designed toward these standards, alongside CE/RoHS, IEC 62368-1 electrical safety, IEC 60529 ingress protection, and EN 15288 pool safety-management practice.

How long does installation take and does the pool need to close?

A typical single-pool deployment takes 3-5 working days. Overhead cameras and the edge unit install without closing the pool. Wall-recessed underwater cameras are best fitted during a scheduled drain-down or maintenance window, while surface-mount underwater kits can be installed by divers without draining. A 2-4 week self-learning period then tunes detection sensitivity to real bather behavior.

Interested in AI Drowning Detection & Aquatic Safety Monitoring System (Overhead + Underwater Vision, Lifeguard Smartwatch Alerts)?

Tell us your scenario and we'll respond with a tailored approach — every project is engineered to your requirements.

Request a Custom Quote

We typically respond within one business day.